November 28, 2012 at 4:29 am
there was an issue yesterday night at the time of rebooting cluster machines after starting machine Windows team raised an issue that SQL server service was not getting started time was around 2 AM, now i am trying to check sql server error logs and default trace of sql server, both are having logs starting from 3:15 AM, i am shocked where all thease logs are gone before 3:15 AM, Has someone can be able to remove log and traces or if it happened, how do we find that some one removed logs or traces???
November 28, 2012 at 6:09 am
How where the machines restarted? A cluster is all about high availabilty, so all nodes in a cluster should not be turned off within the same timeframe if it can be avoided. Before the active node was brought down were the cluster resources moved over to the other node and verified working?
Joie Andrew
"Since 1982"
November 28, 2012 at 6:17 am
yes if the service is offline the files are not locked and can be deleted.
Now how active is the server?
Default trace will only store so much information before it rolls over.
Now for the logs, are you adding in the other logs in the log viewer or are you just viewing the current log?
Have you confirmed that the error logs are not physically on the drive where the error log is configured to live.
November 28, 2012 at 6:58 am
joe,
if suppose some one deleted log when services are offline. then what is the process to make SQL services online again.
November 28, 2012 at 7:04 am
one more thing joe, can you tell me the proper steps to reboot veritas cluster server, one of the guy from windows team directly rebooted active machine then he found that SQL service was not starting and fought with same issue till 2 hrs, and raised ticket to SQL team that issue was related to SQL team only.
November 28, 2012 at 7:11 am
if suppose some one deleted log when services are offline. then what is the process to make SQL services online again.
Which logs are you talking about? Database log files or SQL error logs? Error logs can be deleted and SQL should be able to startup fine. It creates a new one every time SQL is restarted anyway. If database log files are gone you can tell SQL to bring a database online without the log and it will recreate the log, but you will have broken your log chain, so a gap in restore capability is introduced.
What files exactly were deleted?
one more thing joe, can you tell me the proper steps to reboot veritas cluster server, one of the guy from windows team directly rebooted active machine then he found that SQL service was not starting and fought with same issue till 2 hrs, and raised ticket to SQL team that issue was related to SQL team only.
Sorry, can't tell you. I have only worked with Microsoft failover clusters. From a MS cluster, point-of-view, you always check cluster manager to see is SQL is online, because if you are looking at the services MMC on a passive node then SQL is supposed to be offline. I would imagine that Veritas has a similar cluster management tool that you can check the status of SQL on.
Joie Andrew
"Since 1982"
November 28, 2012 at 7:54 am
joe,
well i was talking about Errorl log, i am not sure about anything deleted, only issue is cluster got rebooted yesterdfay and after that SQL service dint come online for 2 hrs in night and reason placed to SQL team, m just troubleshooting and in system log found this error one SQL Server (MSSQLSERVER) service terminated with service-specific error "17058" (0x42A2). and DTC service startup msg just before successful start of SQL service
November 28, 2012 at 8:04 am
So if the cluster was taken offline in the wrong way (i.e. the resources were not brought up on another cluster node and it was not able to failover automatically) then it is possible SQL was down during that 2 hour period until such time it was able to be brought up online. If that is the case then there will be a gap in information in the error log, because SQL was not online writing to it.
If SQL was brought down improperly you should check to ensure that all DBs are online, no errors were reported in the log and you might want to run DBCC CHECKDB commands against your database when you have an available window you can do it in.
Joie Andrew
"Since 1982"
Viewing 8 posts - 1 through 8 (of 8 total)
You must be logged in to reply to this topic. Login to reply