You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Without any hardware error (not server shuntdown, no memory or harddrive problem) we have all the time some index corrupted
* FB 2.5.0.26038 super classic with around 50 000 000 queries / days
* The database is arround 50 GO in size
sometime the index corrupted not block the client application to work, sometime yes because it's make the server enter in undefinite loop.
After Vlad correct the bug with monitoring table the index corrupted problem was not so often BUT it's still happen (before one time every 2/3 days now one time every 2/3 weeks) ... we are still using monitoring table, so it's possible that the problem is still connected to monitoring table
the last time the problem happen, the server become to write in the firebird.log
INET/select_wait: found "not a socket" socket : 504 ... when we try to validate the database we discover that lot of index was corrupted (235 errors founds)
* what do you need exactly to be able to reproduce the bug ? as you see it's involve in a very intensive environement that it's not easy to reproduce
* can the index corrupted appear on "inet error" ?
* can the fb_inet_server can crash without any warning, without writing anything in the log (i thing yes) ? and in this way that can explain the index corrupted if the server crash in the middle of a write ! but for now nothing in the log show that the fbserver crash !
anyway this is a major bug, because index must never be possible to be corrupted without any hardware error ...
thanks for all !
stephane
The text was updated successfully, but these errors were encountered:
Submitted by: vander clock stephane (arkadia)
Votes: 2
Without any hardware error (not server shuntdown, no memory or harddrive problem) we have all the time some index corrupted
* FB 2.5.0.26038 super classic with around 50 000 000 queries / days
* The database is arround 50 GO in size
sometime the index corrupted not block the client application to work, sometime yes because it's make the server enter in undefinite loop.
After Vlad correct the bug with monitoring table the index corrupted problem was not so often BUT it's still happen (before one time every 2/3 days now one time every 2/3 weeks) ... we are still using monitoring table, so it's possible that the problem is still connected to monitoring table
the last time the problem happen, the server become to write in the firebird.log
INET/select_wait: found "not a socket" socket : 504 ... when we try to validate the database we discover that lot of index was corrupted (235 errors founds)
* what do you need exactly to be able to reproduce the bug ? as you see it's involve in a very intensive environement that it's not easy to reproduce
* can the index corrupted appear on "inet error" ?
* can the fb_inet_server can crash without any warning, without writing anything in the log (i thing yes) ? and in this way that can explain the index corrupted if the server crash in the middle of a write ! but for now nothing in the log show that the fbserver crash !
anyway this is a major bug, because index must never be possible to be corrupted without any hardware error ...
thanks for all !
stephane
The text was updated successfully, but these errors were encountered: