Search Results

Search found 2834 results on 114 pages for 'filesystem corruption'.

Page 94/114 | < Previous Page | 90 91 92 93 94 95 96 97 98 99 100 101  | Next Page >

• Permission denied when running Rails app in VirtualBox Ubuntu guest with files on Windows host

  - by Ola Tuvesson

  I think I'm close to having my dev environment set up exactly the way I want, but one final snag remains. I'm running VirtualBox on a Windows 7 64bit host, with my dev enviroment inside a Ubuntu 12.04 guest. I want to keep the files for my projects on the host filesystem - partly so I can access them when the Ubuntu guest is not running, but also so I can use Tortoise and other Windows based tools (cough Photoshop), and it also eases my backup scheme somewhat. So I've got a folder "Rails" on my NTFS drive, which I've shared (Samba) from the host with a user specifically created for the Ubuntu guest. The mount point has been set up and an entry added to fstab (cifs), using a credentials file and the options iocharset=utf8,mode=0777,dir_mode=07??77 This mounts fine and my Ubuntu user has both read and write permissions to the contents. But when I try to start my Rails app I get permission errors on any files the app needs to write to (e.g. the log file) - why is that? Are there any major conceptual flaws with this approach? Would I be better off using the VBox "shared folders" function?

  Read the article

• SSD - Tweaks/Recommended Configuration

  - by Miky D

  I've just purchased my first SSD drive (a 32GB MLC from Imation) without doing enough research ahead of time in the spirit of giving the new technology a shot and getting myself up to speed by empirical research rather than reading countless reviews and I'm now at a crossroads. I've built a new server to test the new drive and at first I wanted to test it with Windows Server 2003 R2 x86 but after I loaded the OS on it and it had problems loading the drivers of the motherboard I went to the internet and did more research and the more I read the more I got confused. Finally I decided to try out Windows Server 2008 R2 x64 since it supposedly has certain support for SSD drives inherent in the NT 6.1 core. Indeed I've had much better luck with the new OS and got all the drivers installed but now I still have some questions: Should I set the drive to: IDE Emulation or AHCI in the BIOS? Should I make any other changes in the BIOS (I've read on the internet that Write Through should be changed to Write Back) Should I make any other adjustments in Windows (i.e. Tweaks such as disabling prefetching or disabling the Last Accessed Timestamp on the filesystem) and if so, is there a good/reliable online resource with instructions? I'm so tired of reading through countless online posts which spend 80% of coverage on the history of SSDs and benchmarks and explanations of how SSDs work. I got that, now I'd like to know if there's anything I should actually do to make sure Windows Server 2008 R2 makes good use of the SSD.

  Read the article

• How to find cause of main file system going to read only mode

  - by user606521

  Ubuntu 12.04 File system goes to readonly mode frequently. First of all I have read this question file system is going into read only mode frequently already. But I have to know if it's not caused by something else than dying hard drive. This is server provided by my client and I am just runing there some node.js workers + one node.js server and I am using mongodb. From time to time (every 20-50h) system suddenly makes filesystem read only, mongodb process fails (due read-only fs) and my node workers/server (which are started by forever) are just killed. Here is the log from dmesg - I can see there some errors and messages that FS is going to read-only, and there is also some JOURNAL error but I would like to find cause of those errors.. http://speedy.sh/Ux2VV/dmesg.log.txt edit smartctl -t long /dev/sda smartctl 5.41 2011-06-09 r3365 [x86_64-linux-3.5.0-23-generic] (local build) Copyright (C) 2002-11 by Bruce Allen, http://smartmontools.sourceforge.net SMART support is: Unavailable - device lacks SMART capability. A mandatory SMART command failed: exiting. To continue, add one or more '-T permissive' options. What I am doing wrong? Same is for sda2. Morover now when I type any command that not exists in shell I get this: Sorry, command-not-found has crashed! Please file a bug report at: https://bugs.launchpad.net/command-not-found/+filebug Please include the following information with the report:

  Read the article

• How can I cache a Subversion password on a server, without storing it in unencrypted form?

  - by Zilk

  My Subversion server only provides access via HTTPS; support for svn+ssh has been dropped because we wanted to avoid creating system users on that machine just for SVN access. Now I'm trying to provide a way for users to cache their passwords for a while, without leaving them stored on the filesystem in unencrypted form. This is no problem for Gnome or KDE users, because they can use gnome-keyring and kwallet, respectively. IIRC, TortoiseSVN has a similar caching mechanism, too. But what about users on a non-GUI system? Some context: in this case, we have a development/testing server where one project has been checked out into the Apache htdocs directory. Development for this project is almost complete, and only minor text/layout changes are performed directly on this server. Nevertheless, the changes should be checked into the repository. There's no kwallet and no gnome-keyring on this system, and the ssh-agent can't help because the repository is accessed via https instead of svn+ssh. As far as I know, that leaves them the choice of entering the password every time they talk to the SVN server, or storing it in an insecure way. Is there any way to get something like what gnome-keyring and kwallet provide in a non-GUI environment?

  Read the article

• Why is mkfs overwriting the LUKS encryption header on LVM on RAID partitions on Ubuntu 12.04?

  - by Starchy

  I'm trying to setup a couple of LUKS-encrypted partitions to be mounted after boot-time on a new Ubuntu server which was installed with LVM on top of software RAID. After running cryptsetup luksFormat, the LUKS header is clearly visible on the volume. After running any flavor of mkfs, the header is overwritten (which does not happen on other systems that were setup without LVM), and cryptsetup will no longer recognize the device as a LUKS device. # cryptsetup -y --cipher aes-cbc-essiv:sha256 --key-size 256 luksFormat /dev/dm-1 WARNING! ======== This will overwrite data on /dev/dm-1 irrevocably. Are you sure? (Type uppercase yes): YES Enter LUKS passphrase: Verify passphrase: # hexdump -C /dev/dm-1|head -n5 00000000 4c 55 4b 53 ba be 00 01 61 65 73 00 00 00 00 00 |LUKS....aes.....| 00000010 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 |................| 00000020 00 00 00 00 00 00 00 00 63 62 63 2d 65 73 73 69 |........cbc-essi| 00000030 76 3a 73 68 61 32 35 36 00 00 00 00 00 00 00 00 |v:sha256........| 00000040 00 00 00 00 00 00 00 00 73 68 61 31 00 00 00 00 |........sha1....| # cryptsetup luksOpen /dev/dm-1 web2-var # mkfs.ext4 /dev/mapper/web2-var [..snip..] Creating journal (32768 blocks): done Writing superblocks and filesystem accounting information: done # hexdump -C /dev/dm-1|head -n5 # cryptsetup luksClose /dev/mapper/web2-var 00000000 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 |................| * 00000400 00 40 5d 00 00 88 74 01 66 a0 12 00 17 f2 6d 01 |.@]...t.f.....m.| 00000410 f5 3f 5d 00 00 00 00 00 02 00 00 00 02 00 00 00 |.?].............| 00000420 00 80 00 00 00 80 00 00 00 20 00 00 00 00 00 00 |......... ......| # cryptsetup luksOpen /dev/dm-1 web2-var Device /dev/dm-1 is not a valid LUKS device. I have also tried mkfs.ext2 with the same result. Based on setups I've done successfully on Debian and Ubuntu (but not LVM or Ubuntu 12.04), it's hard to see why this is failing.

  Read the article

• A complicated nginx/php-fpm chroot setup

  - by Rsaesha

  I'm running nginx and php-fpm, and I want to set up jails for each host. My setup is a little complicated, so following tutorials on the web gets me nowhere. Each site has a directory /var/www/domain.name/ Inside that directory, there will be a public/ directory which will be the website root, a logs/ directory which will store nginx logs for that site specifically, and the chroot filesystem (etc/, usr/, etc.) The first problem I've run into is that nomatter how I configure it, PHP-FPM cannot find the files that are passed to it via nginx. They result in a "Primary script unknown" error, and to make matters worse, the error messages from PHP-FPM are no more verbose than that, so I can't figure out what path is being passed by nginx. A php-fpm pool configuration for a host looks like this: [host] user = host group = www-data chroot = /var/www/domain.name chdir = /public listen = 127.0.0.1:900x 'x' is incremented for each pool. The nginx config for this host looks like this: server { listen 80; server_name domain.name *.domain.name; root /var/www/domain.name/public; index index.php index.html index.html; location ~ \.php$ { expires epoch; fastcgi_split_path_info ^(.+\.php)(/.+)$; include fastcgi_params; fastcgi_index index.php; fastcgi_param SCRIPT_FILENAME $document_root$fastcgi_script_name; fastcgi_pass 127.0.0.1:9001; } } I'm guessing that the problem is the SCRIPT_FILENAME parameter, but I've changed it to just $fastcgi_script_name, and various other combinations, but to no avail. Can anyone help?

  Read the article

• Can't find disk usage in one directory

  - by Xster

  Similar questions are asked frequently but no suggested answers solved my issue. I have some disk space usage that I can't find as well. In df Filesystem 1K-blocks Used Available Use% Mounted on /dev/sda1 144183992 136857180 2652 100% / udev 2013316 4 2013312 1% /dev tmpfs 808848 876 807972 1% /run none 5120 0 5120 0% /run/lock none 2022116 76 2022040 1% /run/shm overflow 1024 0 1024 0% /tmp I checked the inodes, I checked lsof for +L1 or deleted files, I rebooted, I checked for files hidden behind mounts but none of them were the issue. It grows periodically and I'm running out of things to delete to feed the beast. It's all in the home directory of the only user I have. In du in ~ du -h --max-depth=1 192K ./.nv 2.1M ./.gconf 12K ./Pictures 1.6M ./.launchpadlib 12K ./Public 24K ./.TemporaryItems 8.9M ./.cache 12K ./Network Trash Folder 28K ./.vnc 11M ./.AppleDB 48K ./.subversion 1.9G ./.xbmc 8.0K ./.AppleDesktop 12K ./.dbus 81M ./.mozilla 12K ./Music 160K ./.gnome2 44K ./Downloads 692K ./.zsh 236K ./.AppleDouble 64K ./.pulse 4.0K ./.gvfs 1.4M ./.adobe 44K ./.pki 44K ./.compiz-1 168K ./.config 1.4M ./.thumbnails 12K ./Templates 912K ./.gstreamer-0.10 8.0K ./.emacs.d 92K ./Desktop 1.3M ./.local 12K ./Ubuntu One 12K ./Documents 296K ./.fontconfig 12K ./.qt 12K ./.gnome2_private 20K ./.ssh 20K ./.mission-control 12K ./Videos 12K ./Temporary Items 640K ./.macromedia 124G . I can't find a way to figure out how it got to that 124G in that directory. There are no mount points in home.

  Read the article

• DisableCrossAccountCopy not working on some Outlook installs, working on others, both going against Exchange

  - by MikeBaz

  As part of a mail migration project from one Exchange organization to another, we need to be able to prevent users from moving/copying messages between their accounts in each organization. (Yes, users will think this is evil; no, it's not my decision; yes, users will hate us.) Luckily, we thought, Outlook 2010 provides the DisableCrossAccountCopy registry value/policy (cf. http://technet.microsoft.com/en-us/library/ff800883.aspx). (Because you can't do multiple Exchange organizations in a single profile before Outlook 2010, this only matters on Outlook 2010. Yes, I'm ignoring for the sake of this question copy/move to/from the filesystem.) In our test lab, in a test forest with a test Exchange organization, with a second Exchange account added to the profile in either of the "real" Exchange organizations, with the value set to "*", everything works as expected. On a workstation in one of the production domains, however, the setting does not seem to work. We have tried it under HKCU, HKLM, HKCU\Software\Policies, and HKLM\Software\Policies. It simply seems to be ignored. The value was set in the OCT on a test machine, but the OCT (and the ADM/ADMX file) have the wrong type for the value. We have located the value in the registry and removed it everywhere it is found, we think, and put it back in HKCU, but it still isn't taking. At the moment, a clean Outlook install is not an option - even if it was, we at this point would need to know what to do to fix the pushed copy (I didn't push the copy out to thousands of machines, I've just been asked to help clean up the current mess). Thoughts?

  Read the article

• How can ICS in Windows 7 be managed via command line, scripts, config files, etc.?

  - by Skya

  I've been using ICS successfully for years, but now I'm looking for a way to control it through something else than the GUI in Control Panel\Network and Internet\Network Connections - Connection Properties: I want to do everything that the encircled checkbox does, without touching the GUI. But what does the checkbox do? Microsoft don't provide specific information and the most helpful forum post I've found is from 2003. Assuming that some of the advice is still valid, I've come to the conclusion that ICS is broken down into 6 parts that have to be set up individually: the sharedAccess service interface settings firewall rules a static route dnsproxy autodhcp I've already learned that the service can be started/stopped with the command net start/stop sharedAccess and that netsh is a good tool for changing the interface settings and the firewall rules. But I don't understand how ICS handles routing and DNS. All hosts in my network are configured statically, so I don't care much about autodhcp. Thanks for your help! EDIT: I've spent the whole day scanning through ProcMon and I've seen reads/writes to both the registry and the filesystem and it is difficult to determine what parts of it actually make ICS work. I'm trying to look for an API instead. I'm looking into this right now, but I still want to know more about the inner workings.

  Read the article

• Apache doesn't immediately notice a change in the document root

  - by Tom

  We use capistrano for website deployments and our Apache document root is a symlink to a particular code release. The deployment procedure switches the symlink from the old release to the new release as the final step of the deployment. We are migrating our webservers from real servers running RHEL 5.6 to Amazon EC2 virtual machines running Ubuntu 11.10 and the new servers are suffering from a problem where Apache doesn't immediately notice the change to it's document root when the symlink is switched. It can take a second or so (and I think I've even seen it take a couple of minutes). It's kind of like Apache has cached the physical path of the symlink for some time. Does anyone know some Apache settings I could look at to get it to "scan" for changes to it's served files quicker. Thoughts: I read that the disks on virtual machines are much slower (since they are network attached storage). Perhaps the filesystem cache somehow works differently too? If so, is there anything that can be done? The website runs PHP code. Perhaps there is some PHP config differences between RHEL and Ubuntu? I checked realpath_cache_ttl but both servers have it commented out: e.g. ; Duration of time, in seconds for which to cache realpath information for a given ; file or directory. For systems with rarely changing files, consider increasing this ; value. ; http://www.php.net/manual/en/ini.core.php#ini.realpath-cache-ttl ;realpath_cache_ttl = 120 We do use the APC opcode cache but don't think it's the issue due to experimentation. The PHP code is in different file paths for each deployment and we ensure stat=1. Here is a similar question that is very interesting: 294107 - but doesn't provide an answer for me. One solution would be to reload Apache everytime we modify the document root symlink. I'll do this if we can't find another solution.

  Read the article

• postgresql No space left on device

  - by pstanton

  Postgres is reporting that it is out of disk space while performing a rather large aggregation query: Caused by: org.postgresql.util.PSQLException: ERROR: could not write block 31840050 of temporary file: No space left on device at org.postgresql.core.v3.QueryExecutorImpl.receiveErrorResponse(QueryExecutorImpl.java:1592) at org.postgresql.core.v3.QueryExecutorImpl.processResults(QueryExecutorImpl.java:1327) at org.postgresql.core.v3.QueryExecutorImpl.execute(QueryExecutorImpl.java:192) at org.postgresql.jdbc2.AbstractJdbc2Statement.execute(AbstractJdbc2Statement.java:451) at org.postgresql.jdbc2.AbstractJdbc2Statement.executeWithFlags(AbstractJdbc2Statement.java:350) at org.postgresql.jdbc2.AbstractJdbc2Statement.executeUpdate(AbstractJdbc2Statement.java:304) at org.hibernate.engine.query.NativeSQLQueryPlan.performExecuteUpdate(NativeSQLQueryPlan.java:189) ... 8 more However the disk has quite a lot of space: Filesystem Size Used Avail Use% Mounted on /dev/sda1 386G 123G 243G 34% / udev 5.9G 172K 5.9G 1% /dev none 5.9G 0 5.9G 0% /dev/shm none 5.9G 628K 5.9G 1% /var/run none 5.9G 0 5.9G 0% /var/lock none 5.9G 0 5.9G 0% /lib/init/rw The query is doing the following: INSERT INTO summary_table SELECT t.a, t.b, SUM(t.c) AS c, COUNT(t.*) AS count, t.d, t.e, DATE_TRUNC('month', t.start) AS month, tt.type AS type, FALSE, tt.duration FROM detail_table_1 t, detail_table_2 tt WHERE t.trid=tt.id AND tt.type='a' AND DATE_PART('hour', t.start AT TIME ZONE 'Australia/Sydney' AT TIME ZONE 'America/New_York')>=23 OR DATE_PART('hour', t.start AT TIME ZONE 'Australia/Sydney' AT TIME ZONE 'America/New_York')<13 GROUP BY month, type, t.a, t.b, t.d, t.e, FALSE, tt.duration any tips?

  Read the article

• On a failing hard drive, I am able to view data but unable to copy it - why?

  - by Tom

  I have a 2.5" external hard drive that is failing. It's not making the expected 'clicking' noise that most hard drives and I am able to view the data, but I am unable to actually retrieve the data. I attempted to use SpinRite in order to access the data on the drive, but it didn't like the external drive. When I view the drive's property page, the drive shows that it's used space is at 100% and that it has 0 bytes available; however, the progress indicator under the drive icon in Windows Explorer shows that it's roughly 50% full (which is correct). When I attempt to run Windows' "Error Checking" tool and attempt to "scan for an attempt recovery of bad sectors," the tool begins to run then immediately closes with no error message. I am able to browse the contents of the drive using Windows Explorer. When I begin to try copying any given single file, the copy process begins, an indicator starts, and then the copy fails with no real error message. The Disk Management page in Computer Management under Control Panel also shows this drive has being 'Healthy.' I dropped the drive off at a data recovery store and they said that "The data seems to be intact, but an internal failure is preventing any information from being retrieved." They offered to provide me references to a data recovery specialist. I've also attempted to run CHKDSK on the drive (with and without arguments) but it returns the following error: The type of the filesystem is RAW. CHKDSK is not available for RAW drives. Before going the route of more expensive data recovery, I'm wondering if these symptoms sound familiar to anyone? Other questions... I'm willing to continue trying tools such as TestDisk and/or PhotoRec (as the majority of the data that I'd like to salvage are photos) but how long I should expect either tool to run given approximately 400GB of data? I'm also comfortable using Linux so I welcome any suggestions for utilities or tools and strategies with which you've had success.

  Read the article

• NTFS: Deny all permissions for all files, except where explicitly added

  - by Simon

  I'm running a sandboxed application as a local user. I now want to deny almost all file system permissions for this user to secure the system, except for a few working folders and some system DLLs (I'll call this set of files & directories X below). The sandbox user is not in any group. So it shouldn't have any permissions, right? Wrong, because all "Authenticated Users" are a member of the local "Users" group, and that group has access to almost everything. I thought about recursively adding deny ACL-entries to all files and directories and remove them manually from X. But this seems excessive. I also thought about removing "Authenticated Users" from the "Users" group. But I'm afraid of unintended side-effects. It's likely that other things rely on this. Is this correct? Are there better ways to do this? How would you limit the filesystem permissions of a (very) non-trustworthy account?

  Read the article

• Fedora 9 not reconizing hard drive

  - by Andrew Jones

  I am installing Fedora 9 to a PC (specifications at the bottom) and have had a lot of trouble with it recognising the hard drive. To get the Fedora installer to recognize it in the first place I had to pass "ata_generic.all_generic_ide=1 pci=nomsi" to the kernel, after which it installed OK. However, now when I boot the installed OS, I get a "could not find filesystem '/dev/root'" error. I tried passing the same arguments to the kernel at boot as I did when installing but to no avail. I have tried using the default LVM layout and defining manual ones but it made no difference. There is no option in the BIOS to enable AHCI or anything like that, in fact the BIOS is very limited in most respects. I can get into the system by using the installation CD in rescue mode (with those extra kernal parameters) but I'm not sure what to do once in there... Unfortunately using a more recent version of Fedora or even another Linux distribution altogether isn't an option becuase of outside constraints - which is annoying since I know for a fact Ubuntu works fine on this setup. I have not been using Linux that long, so treat me like an idiot - I am one. Any help would be greatly appreciated, thanks! System spec: Intel Atom Z530 CPU @ 1.6 GHz Intel US15W chipset 1 GB DDR2 160 GB SATA harddisk (Samsung HM16HI) 1000 Mbit/s Ethernet port Phoenix BIOS

  Read the article

• EC2 Configuration

  - by user123683

  I am trying to create a server structure for my EC2 account. The design I have chosen consists of 2 instances running in different availability zones, elastic load balancer, an auto-scaling group with cloudwatch monitoring configured and a security group defining rules for access to the instances. This setup is to support an online web application written in PHP. I am trying to decide what is a better policy: Store MySQL DB on a separate Instance Store MySQL DB on an attached EBS volume (from what i know auto-scaling will not replicate the attached EBS volume but will generate new instances from a chosen AMI - is this view correct?) Regards the AMI I plan to use a basic Amazon linux 64 bit AMI, and install bastille (maybe OSSEC) but I am looking to also use an encrypted file system. Are there any issues using an encrypted file system and communication between the DB and webapp i neeed to be aware of? Are there any comms issues using the encrypted filesystem on the instance housing the webapp I was going to launch a second instance or attach a second volume in the second availability zone to act as a standby for the database - I'm just looking for some suggestions about how to get the two DB's to talk - will this be a big task Regards updates for security is it best to create a recent snapshot and just relaunch and allow Amazon to install updates on launch or is the yum update mechanism a suitable alternative - is it better practice to relaunch instead of updates being installed which force a restart. I plan to create two AMI snapshots one for the app server and one for the DB each with the same security measures in place - is this a reasonable - I just figure it is a better policy than having additional applications that are unnecessary included in a AMI that I intend on using. My plan for backup is to create periodic snapshots of the webapp and DB instances (if I use an additional EBS volume instead of separate instances my understanding is that the EBS volume will persist in S3 storage in the event of an unexpected termination and I can create snapshots of the volume backup purposes). Thanks in advance for suggestions and advice. I am new to EC2 and I may have described unnecessary overkill but I want to try implement what can be considered a best practice solution so all advice is appreciated.

  Read the article

• Nginx's speed, and how to replicate it [migrated]

  - by Mediocre Gopher

  I'm interested in this from more than an academic standpoint rather than a practical standpoint; I don't plan on creating a production webserver to compete with nginx. What I'm wondering is how exactly nginx is so fast. The top google response for this is this thread, but it merely links to a cryptic slideshow and a general covering of different io strategies. All other results seem to simply describe how fast nginx is, rather then the reason. I tried building a simple erlang server to try to compete with nginx, but to no avail; nginx won out. All my server does is spawn a new process for each request, uses that process to read the file to a socket, then closes the file and kills the thread. It's not complicated, but given erlang's lightweight processes and underlying aio structure I thought it would compete, but nginx still wins out by a consistent 300 ms average under a heavy stress test. What is nginx doing that my simple server isn't? My first thought would be keeping files in main memory instead of tossing them between requests, but the filesystem cache does this already so I didn't think it would make that great of difference. Am I wrong? Or is there something else that I'm missing?

  Read the article

• DBCC CHECKDB (BatmanDb, REPAIR_ALLOW_DATA_LOSS) &ndash; Are you Feeling Lucky?

  - by David Totzke

  I’m currently working for a client on a PowerBuilder to WPF migration.  It’s one of those “I could tell you, but I’d have to kill you” kind of clients and the quick-lime pits are currently occupied by the EMC tech…but I’ve said too much already. At approximately 3 or 4 pm that day users of the Batman[1] application here in Gotham[1] started to experience problems accessing the application.  Batman[2] is a document management system here that also integrates with the ERP system.  Very little goes on here that doesn’t involve Batman in some way.  The errors being received seemed to point to network issues (TCP protocol error, connection forcibly closed by the remote host etc…) but the real issue was much more insidious. Connecting to the database via SSMS and performing selects on certain tables underlying the application areas that were having problems started to reveal the issue.  You couldn’t do a SELECT * FROM MyTable without it bombing and giving the same error noted above.  A run of DBCC CHECKDB revealed 14 tables with corruption.  One of the tables with issues was the Document table.  Pretty central to a “document management” system.  Information was obtained from IT that a single drive in the SAN went bad in the night.  A new drive was in place and was working fine.  The partition that held the Batman database is configured for RAID Level 5 so a single drive failure shouldn’t have caused any trouble and yet, the database is corrupted.  They do hourly incremental backups here so the first thing done was to try a restore.  A restore of the most recent backup failed so they worked backwards until they hit a good point.  This successful restore was for a backup at 3AM – a full day behind.  This time also roughly corresponds with the time the SAN started to report the drive failure.  The plot thickens… I got my hands on the output from DBCC CHECKDB and noticed a pattern.  What’s sad is that nobody that should have noticed the pattern in the DBCC output did notice.  There was a rush to do things to try and recover the data before anybody really understood what was wrong with it in the first place.  Cooler heads must prevail in these circumstances and some investigation should be done and a plan of action laid out or you could end up making things worse[3].  DBCC CHECKDB also told us that: repair_allow_data_loss is the minimum repair level for the errors found by DBCC CHECKDB Yikes.  That means that the database is so messed up that you’re definitely going to lose some stuff when you repair it to get it back to a consistent state.  All the more reason to do a little more investigation into the problem.  Rescuing this database is preferable to having to export all of the data possible from this database into a new one.  This is a fifteen year old application with about seven hundred tables.  There are TRIGGERS everywhere not to mention the referential integrity constraints to deal with.  Only fourteen of the tables have an issue.  We have a good backup that is missing the last 24 hours of business which means we could have a “do-over” of yesterday but that’s not a very palatable option either. All of the affected tables had TEXT columns and all of the errors were about LOB data types and orphaned off-row data which basically means TEXT, IMAGE or NTEXT columns.  If we did a SELECT on an affected table and excluded those columns, we got all of the rows.  We exported that data into a separate database.  Things are looking up.  Working on a copy of the production database we then ran DBCC CHECKDB with REPAIR_ALLOW_DATA_LOSS and that “fixed” everything up.   The allow data loss option will delete the bad rows.  This isn’t too horrible as we have all of those rows minus the text fields from out earlier export.  Now I could LEFT JOIN to the exported data to find the missing rows and INSERT them minus the TEXT column data. We had the restored data from the good 3AM backup that we could now JOIN to and, with fingers crossed, recover the missing TEXT column information.  We got lucky in that all of the affected rows were old and in the end we didn’t lose anything.  :O  All of the row counts along the way worked out and it looks like we dodged a major bullet here. We’ve heard back from EMC and it turns out the SAN firmware that they were running here is apparently buggy.  This thing is only a couple of months old.  Grrr…. They dispatched a technician that night to come and update it .  That explains why RAID didn’t save us. All-in-all this could have been a lot worse.  Given the root cause here, they basically won the lottery in not losing anything. Here are a few links to some helpful posts on the SQL Server Engine blog.  I love the title of the first one: Which part of 'REPAIR_ALLOW_DATA_LOSS' isn't clear? CHECKDB (Part 8): Can repair fix everything? (in fact, read the whole series) Ta da! Emergency mode repair (we didn’t have to resort to this one thank goodness)   Dave Just because I can…   [1] Names have been changed to protect the guilty. [2] I'm Batman. [3] And if I'm the coolest head in the room, you've got even bigger problems...

  Read the article

• btrfs: can i create a btrfs file system with data as jbod and metadata mirrored

  - by Yogi

  I am trying to build a home server that will be my NAS/Media server as well a the XBMC front end. I am planning on using Ubuntu with btrfs for the NAS part of it. The current setup consists of 1TB hdd for the OS etc and two 2TB hdd's for data. I plan to have the 2TB hdd's used as JBOD btrfs system in which i can add hdd's as needed later, basically growing the filesystem online. They way I had setup the file system for testing was while installing the OS just have one of the HDD's connected and have btrfs on it mounted as /data. Later on add another hdd to this file system. When the second disk was added btrfs made as RAID 0, with metadata being RAID 1. However, this presents a problem: even if one of the disk fails I loose all my data (mostly media). Also most of the time the server will be running without doing any disk access, i.e. the HDD's can be spun down, when a access request comes in this with the current RAID 0 setup both disks will spin up. in case I manage a JBOD only the disk that has the file needs to be spun up. This should hopefully reduce the MTBF for each disk. So, is there a way in which I can have btrfs setup such that metadata is mirrored but data stays in a JBOD formation? Another question I have is this, I understand that a full drive failure in JBOD will lose data on the drive, but having metadeta mirrored across all drives, will this help the filesytem correct errors that migh creep in (ex bit rot?) and is btrfs capable of doing this.

  Read the article

• Data recovery on a corrupted 3TB disk

  - by Mark K Cowan

  Short version I probably need software to run a deep-scan recovery (ideally on Linux) to find files on NTFS filesystem. The file data is intact, but the references are no longer present. Analogous to recovering data from a "quick-formatted" partition. Hopefully there is a smarter way available than deep-scan, one which would recover filenames and possibly paths. Long version I have a 3TB disk containing a load of backups. Windows 7 SP1 refused to detect the disk when plugged in directly via SATA, so I put it on a USB/SATA adaptor which seemed to work at first. The SATA/USB adaptor probably does not support disks over 2.2TB though. Windows first asked me if I wanted to 'format' the disk, then later showed me most of the contents but some folder were inaccessible. I stupidly decided to run a CHKDSK on my backup disk, which made the folders accessible but also left them empty. I connected this disk via SATA to my main PC (Arch Linux). I tried: testdisk ntfsundelete ntfsfix --no-action (to look for diagnostically relevant faults, disk was "OK" though) to no avail as the files references in the tables had presumably been zeroed out by CHKDSK, rather than using a typical journal'd deletion). If it is useful at all, a majority of the files that I want to recover are JPEG, Photoshop PSD, and MPEG-3/MPEG-4/AVI/MKV files. If worst comes to worst, I'll just design my own sector scanner and use some simple heuristic-driven analysis to recover raw binary blocks of data from the disk which appears to match the structures of the above file types. I am unfamiliar with the exact workings of NTFS but used to be proficient at recovering FAT32 systems with just a hex-editor, so I can provide any useful diagnostic information if you let me know how to find it! My priorities in ascending order of importance for choosing the accepted answer: Restores directory structure Recovers many filenames in addition to the file data Is free / very cheap Runs on Linux Recovers a majority of file data The last point is the most important, but the more of the higher points you match the more rep you'll probably get :)

  Read the article

• Are relative-path symlinks reliable on Rackspace Cloud Sites?

  - by Jakobud

  Rackspace's Cloud Sites have a lot of stupid limitations. For example, no SSH (in or out), no shell, no RSYNC, etc... (even through cron). Recently I learned that you can't reliably use symlinks in Cloud Sites. Apparently this is because the absolute path of your sites could change at any moment, since it's a shared host environment split up between many disks/servers. I guess different account's sites get moved from disk to disk whenever Rackspace decides to. Supposedly to increase efficiency across the board. So after talking with a Rackspace tech, he said they cannot guarantee that symlinks would always work. Obviously this is because if you have a symlink that use's an absolute path like this: //mnt/disk-34566/home/user34566/files/sites/www.mysite.com/mydir If you files go moved to a different disk (or whatever they do), then the absolute path would be different and the link would now be broken. That makes sense. So next, I asked the Rackspace tech if relative path symlinks were reliable. So if I have the following link: files/sites/www.mysite.com/mylink --> ../www.myothersite.com/anotherdir You can see that the symlink simply points to a nearby directory's sub-directory. He said they cannot guarantee that even those would always work either. Since it uses a relative path to another nearby directory I'm not sure how it could ever break from something Rackspace would do. Do relative symlinks somehow rely on absolute paths underneath? Or is Rackspace using some weird custom filesystem where they will break from absolute path changes? It seems like a relative-path symlink would be fine and would only break if the user did something to mess up the directories involved. But when the tech's say that they "don't officially support symlinks of any kind" that makes me hesitant to use them for large commercial websites in Cloud Sites. Can anyone with Rackspace experience give input on this topic?

  Read the article

• Burn bootable iso image to USB stick using dd: Won't boot (despite USB first in boot sequence)

  - by Nicolas Raoul

  I have installed Ubuntu on a Lenovo Thinkpad R500 2732, and I must update the BIOS. On the Lenovo website, I am offered this: BIOS Update Bootable CD for Windows 7 (32-bit, 64-bit), Vista (32-bit, 64-bit), XP - ThinkPad R500 I guess a bootable CD that would do a BIOS update is indeed what I need. (still wondering why it says "Windows" though... if it is bootable should not it be OS-agnostic?) Not wanting to waste a CD, I copied the image to my USB stick: sudo dd if=/home/nico/7yuj40uc.iso of=/dev/sdb1 bs=1M And rebooted, after making sure USB is first in the boot sequence. PROBLEM: It does not boot. Did I forget one step? Details about the iso image (readme): ls -lh 7yuj40uc.iso 25M file 7yuj40uc.iso /home/nico/7yuj40uc.iso: # ISO 9660 CD-ROM filesystem data '7YUJ40US ' (bootable) (Scroll to the right: it says "bootable") UNetbootin does not work because it is not a Linux image. Some people on the Internet advise to copy the content of the ISO and do other steps. This ISO has zero ISO content so it would not work. If I mount the ISO, I can see it contains zero files.

  Read the article

• Matlab computations done over Apple Filing Protocol (AFP) depend on POSIX permissions, ignores ACLs

  - by flumignan

  I'm a system administrator and have never used Matlab, so forgive my general ignorance of the program. My users have encountered problems when executing scripted Matlab actions over AFP to a Mac OS X Server 10.6.7 where the access control list (ACL) should allow actions, but the POSIX-style permissions disallow the activity. It seems as if Matlab, run locally on the Mac workstations on datasets on the remote server, ignores the ACLs entirely. This is the only application I've ever seen behave this way. The server's filesystem is HFS+J and all other activity is performing as expected. These users cannot use CIFS because of our integration with external directory systems. In this example, the directory bxdata, the members of the group cibturner should be able to modify the files. Indeed, they can using any other method except via Matlab scripts. When the Matlab script hits these files, the POSIX permissions of 644 disallow modification. It's as if the ACLs are irrelevant. [root@cib 16:00:24 /14181.2_5sM]# ls -leh@ bxdata/ total 128 -rw-r--r--+ 1 kel32 staff 18K Feb 15 09:31 TS-5sMath030708-21073-1.edat 0: group:cibturner inherited allow read,write,execute,append,readattr,writeattr,readextattr,writeextattr,readsecurity,writesecurity,chown 1: group:cibsrlocaladmins inherited allow read,write,execute,append,readattr,writeattr,readextattr,writeextattr,readsecurity,writesecurity,chown 2: group:crcservergroup inherited allow read,write,execute,append,readattr,writeattr,readextattr,writeextattr,readsecurity,writesecurity,chown -rw-r--r--+ 1 kel32 staff 25K Feb 15 09:31 TS-5sMath030708-21073-1.txt 0: group:cibturner inherited allow read,write,execute,append,readattr,writeattr,readextattr,writeextattr,readsecurity,writesecurity,chown 1: group:cibsrlocaladmins inherited allow read,write,execute,append,readattr,writeattr,readextattr,writeextattr,readsecurity,writesecurity,chown 2: group:crcservergroup inherited allow read,write,execute,append,readattr,writeattr,readextattr,writeextattr,readsecurity,writesecurity,chown Because this server has HIPAA data, security is critical. We are not using networked home directories or SAN technology. The MatLab program is run on the user's hard drive; access is granted via Kerberized AFP.

  Read the article

• bind9 dlz/mysql at ubuntu segfault libmysqlclient.so

  - by Theos

  I have a big problem. I installed the bind9 nameserver to three different computer. two Ubuntu 10.04.4 LTS, and one Ubuntu 11.10 I compiled it 9.7.0, 9.7.3, 9.9.0 with this method: ./configure --prefix=/usr --sysconfdir=/etc/bind --localstatedir=/var \ --mandir=/usr/share/man --infodir=/usr/share/info \ --enable-threads --enable-largefile --with-libtool --enable-shared --enable-static \ --with-openssl=/usr --with-gssapi=/usr --with-gnu-ld \ --with-dlz-mysql=yes --with-dlz-bdb=no \ --with-dlz-filesystem=yes --with-geoip=/usr make make install After the set up for dlz/mysql, the BIND server is working perfetctly until 5-30 minute long. Ahter i got segfault. I resolve temporaly the problem with a simple process watchdog, and if the named is stopped, the watchdog is restart it, but this is not a good idea in long therm. My log output is: messages: Apr 13 19:33:51 dnsvm kernel: [ 8.088696] eth0: link up Apr 13 19:33:58 WATCHDOG: named not running. Restarting Apr 13 19:35:08 dnsvm kernel: [ 87.082572] named[1027]: segfault at 88 ip b71c4291 sp b5adfe30 error 4 in libmysqlclient.so.16.0.0[b714e000+1aa000] Apr 13 19:35:08 WATCHDOG: named not running. Restarting Apr 13 19:35:08 dnsvm kernel: [ 87.457510] named[1423]: segfault at 68 ip b71d6122 sp b52f0a40 error 4 in libmysqlclient.so.16.0.0[b7160000+1aa000] Apr 13 19:35:09 WATCHDOG: named not running. Restarting Apr 13 19:41:56 dnsvm kernel: [ 494.838206] named[1448]: segfault at 88 ip b731c291 sp b5436e30 error 4 in libmysqlclient.so.16.0.0[b72a6000+1aa000] Apr 13 19:41:57 WATCHDOG: named not running. Restarting Apr 13 19:57:26 dnsvm kernel: [ 1424.023409] named[2976]: segfault at 88 ip b72d1291 sp b6beee30 error 4 in libmysqlclient.so.16.0.0[b725b000+1aa000] Apr 13 19:57:26 WATCHDOG: named not running. Restarting Apr 13 20:11:56 dnsvm kernel: [ 2294.324663] named[6441]: segfault at 88 ip b7357291 sp b6473e30 error 4 in libmysqlclient.so.16.0.0[b72e1000+1aa000] Apr 13 20:11:57 WATCHDOG: named not running. Restarting syslog: http://pastebin.com/hjUyt8gN the first server is a native, normal x64 server (u1004lts), the second is virtualised server (u11.10) the third is also virtualised (10.04lts) This servers is only for dns providing with mysql server db. But the problem is be with all server, and all bind version. named.conf: http://pastebin.com/zwm1yP7V Can anybody help me, or any good idea?

  Read the article

• OS X Client & Ubuntu Server - Best way for client to access files on server?

  - by Camsoft

  I've got a local development web server running Ubuntu. I also have an iMac running OS X 10.6 which I use a client and is my development machine. I'm currently have Samba server installed on my Ubuntu server. I have shares setup for all the website directories. I then use my Mac and Coda to edit the files via their shares. This generally works really well but I noticed that my Mac was writing loads of resource fork ._filename files everywhere. I found out the following about the files: These files are created on volumes that don't natively support full HFS file characteristics (e.g. ufs volumes, Windows fileshares, etc). When a Mac file is copied to such a volume, its data fork is stored under the file's regular name, and the additional HFS information (resource fork, type & creator codes, etc) is stored in a second file (in AppleDouble format), with a name that starts with "._". (These files are, of course, invisible as far as OS-X is concerned, but not to other OS's; this can sometimes be annoying...) Does anyone know of a way of sharing files between a Mac client and a Linux server that is most compantable between the two operation systems? Ideally it needs to support the HFS filesystem so that the resource forks are not created and it also needs to support the permissions between server and client.

  Read the article

• Howto: SaaS / PHP Application / Tenants / Security

  - by Ben Fransen

  Hi all, Being completely new in the webhostingcorner I have a few questions on how to implement/setup a webserver for a SaaS application. I'm about to rent my own server for a new product (CMS) I'm launching in two months. Developing the system wasn't that much of wild ride to me, but a correct way to implement it, is. So lets say this is my situation: I want to host 10 websites for 8 clients. There are 6 single sites, and two clients have two websites they can manage with my software. The CMS must be placed on the server too, all clients are connecting to 1 system The database must be placed Depending on the contract a client makes, the client gets some storage. How to measure the used storage over the DB, FileSystem and email Clients may not, in any case be able to somehow get outside their directory, but from the CMS directory the CMS must be able to create files and dirs in a clients directory (for templates, imagegalleries, widgets, etc, etc). I was thinking about something like a dirstructure like this: ./CMS/ [all CMS files] ./Websites/*/ [all websites] My hostingprovider will install updates to the os (CentOS, latest) and the admin panel (Direct Admin). Is there anybody with experience on this topic? Or do you have some thoughts about it? please join the conversation since I'm completely new to this. Ben

  Read the article

< Previous Page | 90 91 92 93 94 95 96 97 98 99 100 101  | Next Page >