Bug #34422 | File system write failed during LogFileOperationRecord state 17. OS errno: 5 | ||
---|---|---|---|
Submitted: | 8 Feb 2008 16:09 | Modified: | 27 Mar 2014 15:40 |
Reporter: | Cyril SCETBON | Email Updates: | |
Status: | Can't repeat | Impact on me: | |
Category: | MySQL Cluster: Cluster (NDB) storage engine | Severity: | S1 (Critical) |
Version: | 5.1.44-ndb-7.1.4b | OS: | Linux (debian etch) |
Assigned to: | CPU Architecture: | Any | |
Tags: | cluster, crash, ndb |
[8 Feb 2008 16:09]
Cyril SCETBON
[8 Feb 2008 16:12]
Cyril SCETBON
ndb_trace files
Attachment: trace.zip (application/x-zip-compressed, text), 350.89 KiB.
[8 Feb 2008 16:21]
Hartmut Holzgraefe
Operating system error code 5 indicates a file system or storage device level error: $ perror 5 OS error code 5: Input/output error so this actually looks more like a system problem than a MySQL bug. To fully analyze this kind of bugs we need the cluster configuration file and the complete cluster log files: - the management nodes ndb_#_cluster.log files - all data nodes ndb_#_error.log files - all data nodes ndb_#_out.log files - all data nodes ndb_#_trace.log.* files and usually also - all mysql nodes error log files Collecting these files can be automated using the NDB error-reporter utility which is part of the cluster tools distribution and documented in our online manual: http://dev.mysql.com/doc/refman/5.1/en/mysql-cluster-utilities-ndb-error-reporter.html
[9 Feb 2008 18:34]
Cyril SCETBON
> so this actually looks more like a system problem than a MySQL bug. That's what we supposed but no error is logged by the OS (dmesg, messages,syslog or others) and the output log files (trace files and others) are on the same device and are written !!! So why did we get the error logged on the device that raised I/O errors ?
[10 Apr 2008 22:13]
Sveta Smirnova
Cyril, thank you for the feedback. Could you please provide requested information: - the management nodes ndb_#_cluster.log files - all data nodes ndb_#_error.log files - all data nodes ndb_#_out.log files - all data nodes ndb_#_trace.log.* files - all mysql nodes error log files
[10 May 2008 23:00]
Bugs System
No feedback was provided for this bug for over a month, so it is being suspended automatically. If you are able to provide the information that was originally requested, please do so and change the status of the bug back to "Open".
[26 Feb 2009 22:05]
Yves Trudeau
I just hit exactly the same bug: Time: Thursday 26 February 2009 - 21:11:35 Status: Temporary error, restart node Message: Temporary on access to file (Internal error, programming error or missi ng error message, please report a bug) Error: 2809 Error data: DBLQH: File system write failed during LogFileOperationRecord state 17. OS errno: 5 Error object: DBLQH (Line: 13086) 0x0000000e Program: /ndb/local/libexec/ndbd Pid: 4348 Trace: /ndb/mysql-cluster/ndb_29_trace.log.1 Version: mysql-5.1.32 ndb-6.3.23-GA ***EOM*** And we solved the issue by setting "ODirect = 0" in the config.ini file. It looks like there is a odd behavior with ODirect = 1. The cluster has 24 ndbd nodes, I don't have all the log files. Yves
[27 Feb 2009 9:18]
Cyril SCETBON
We have not met this bug anymore since we are in 5.1.27-ndb-6.3.17
[4 Jan 2010 14:19]
Cyril SCETBON
Hi we actually hit this bug for the third time. I send you our our logs and configuration file.
[16 Apr 2010 12:56]
Cyril SCETBON
any news ?
[23 Jun 2010 10:22]
Cyril SCETBON
setting ODirect=0 does not solve our issue. Any other idea ?
[23 Jun 2010 10:23]
Cyril SCETBON
version changed
[24 Jun 2010 8:23]
Cyril SCETBON
we can't use the replicated site until this bug is solved !
[2 Jul 2010 13:00]
Cyril SCETBON
we upgraded our slave cluster in our architecture (1 cluster replicated over another one). We still met the same error on a datanode: Time: Friday 2 July 2010 - 10:02:04 Status: Temporary error, restart node Message: Temporary on access to file (Internal error, programming error or missing error message, please report a bug) Error: 2809 Error data: DBLQH: File system write failed during LogFileOperationRecord state 17. OS errno: 5 Error object: DBLQH (Line: 14516) 0x0000000e Program: /usr/sbin/ndbd Pid: 30951 Version: mysql-5.1.44 ndb-7.1.4b Trace: /data/ndb_5_trace.log.2 ***EOM***
[31 Aug 2010 6:08]
Jonas Oreland
Hi, os error 5 is "Input/output error", this is something that we can't/don't handle gracefully... do you have hw errors/bad sectors? what os(kernel) do you run /Jonas
[31 Aug 2010 6:10]
Jonas Oreland
Hi again, I quite sure that another user had this...and it disappeared when they upgrade linux kernel...
[31 Aug 2010 14:44]
Cyril SCETBON
If you have more information it would be appreciated. Each time we met this error there was no information in the system log files. we checked our disks and nothing appeared to be in error.
[14 Sep 2010 6:30]
Jonas Oreland
so what os version do you use ? what filesystem do you use ? /Jonas
[14 Sep 2010 7:45]
Cyril SCETBON
debian/etch + ext3 filesystem
[14 Sep 2010 8:07]
Jonas Oreland
Is upgrading to a newer linux kernel an option? Google said that etch used 2.6.18 an was released around 2007... /Jonas
[14 Sep 2010 8:23]
Cyril SCETBON
not yet :( we're using a 2.6.24 kernel but we'll try it on ubuntu lucid next year. Maybe the error will disappear, but I don't know. No one else met this error ? it's weird cause we only met it with mysql cluster and no other software.
[25 Feb 2014 19:20]
Sveta Smirnova
Thank you for the feedback. Did you update the kernel as was planned? Did it solve the error?
[26 Mar 2014 1:00]
Bugs System
No feedback was provided for this bug for over a month, so it is being suspended automatically. If you are able to provide the information that was originally requested, please do so and change the status of the bug back to "Open".
[26 Mar 2014 8:29]
Cyril Scetbon
Sorry I can't talk about it anymore cause I changed position and was not able to test it again ..
[27 Mar 2014 15:40]
Sveta Smirnova
Thank you for the feedback. I can only set status of this report as "Can't repeat" in this case. If you hit same issue again with newer kernel feel free to reopen it.