Subject | Memory consumption or database corruption |
---|---|
Author | Aldo Caruso |
Post date | 2016-12-23T20:08:49Z |
Hi,
I'm using firebird 2.5 superserver, php5 and apache2 on a Linux server.
I noticed an increase in memory ( 50 MB / day ) in the last months,
that forced me to periodically restart apache server (once every two or
three weeks).
Yesterday, after an apache restart, memory started to increase out
of control in a few minutes, as well as the number of apache processes (
normally 30, yesterday more than 100 ) and the number of firebird
threads ( normally 10, yesterday more than 100 ). Needless to say that
the web page didn't respond.
I tried restarting firebird server, but nothing changed ( memory
and processes started to grow out of control again ).
I took caution measures ( stopped apache, firebird, made a copy of
the database file, etc.)
I stopped apache server, shut down the database, ran "gfix
-validate -full" which threw the error shown below.
Summary of validation errors
Number of record level errors : 1
Also in the /var/log/firebird2.5.log I found the following lines
( when the problem began )
INET/inet_error: read errno = 104
( when doing gfix -validate )
Relation has 2407 orphan backversions (0 in use) in table PROPS (147)
Afterwards I made a back up and restore of the database, no errors
were shown during those processes, and everything went back to normality.
I wonder whether these errors point out some kind of database
corruption and if they are related with the memory consumption symptoms
I described before.
Thanks
Aldo Caruso
I'm using firebird 2.5 superserver, php5 and apache2 on a Linux server.
I noticed an increase in memory ( 50 MB / day ) in the last months,
that forced me to periodically restart apache server (once every two or
three weeks).
Yesterday, after an apache restart, memory started to increase out
of control in a few minutes, as well as the number of apache processes (
normally 30, yesterday more than 100 ) and the number of firebird
threads ( normally 10, yesterday more than 100 ). Needless to say that
the web page didn't respond.
I tried restarting firebird server, but nothing changed ( memory
and processes started to grow out of control again ).
I took caution measures ( stopped apache, firebird, made a copy of
the database file, etc.)
I stopped apache server, shut down the database, ran "gfix
-validate -full" which threw the error shown below.
Summary of validation errors
Number of record level errors : 1
Also in the /var/log/firebird2.5.log I found the following lines
( when the problem began )
INET/inet_error: read errno = 104
( when doing gfix -validate )
Relation has 2407 orphan backversions (0 in use) in table PROPS (147)
Afterwards I made a back up and restore of the database, no errors
were shown during those processes, and everything went back to normality.
I wonder whether these errors point out some kind of database
corruption and if they are related with the memory consumption symptoms
I described before.
Thanks
Aldo Caruso