Subject | SV: [firebird-support] System performance issue |
---|---|
Author | Poul Dige |
Post date | 2012-12-13T10:24:03Z |
Could it be the old "Windows save point" problem that Windows, on file access, makes a complete backup of the firebird file?
Poul Dige
(Sorry for the formatting but Outlook has its own ways...)
Fra: firebird-support@yahoogroups.com [mailto:firebird-support@yahoogroups.com] På vegne af Robert martin
Sendt: 12. december 2012 23:57
Til: firebird-support@yahoogroups.com
Emne: [firebird-support] System performance issue
Hi
One of our clients has a machine where intermittently the system crawls
to almost stop. Launching programs (like paint) takes about 2 minutes
to open. The CPU usage is low (1-3 %) but HDD usage shows at between 97
- 100%. This situation can last 30 mins to 2 hrs before returning to
normal.
I raise the issue here because we have traced the cause to Firebird.
When the machine is on a go slow, stopping the Firebird service fixes
the problem. Sometimes starting the service again and connecting breaks
the system again, sometimes it doesn't. I had though the issue may have
been garbage collection but we have moved from FB 2.0 super server to FB
2.5.2 super classic and are still getting the problem. I'm pretty sure
garbage collection is supposed to be a low priority activity, right?
The system often enters 'slow more' straight after being rebooted.
The weird thing is that while the system is slow and the HDD shows at
around 100% use in resource monitor, the actually data throughput is
about 1 ro 5MB a sec. This is can be much higher when the system runs
normally. This made me think the HDD might be faulty but all
indications are that its not (according to a hardware technician) and of
course the problem fixes itself when stopping Firebird.
System is
i7 quad core (8 cores in Windows)
10 GB Ram
Windows 7 64bit
16GB free hdd space
Firebird 2.5.2 64bit
DB size is 2.7 GB
Any suggestions would be greatly appreciated. I know that there are
monitoring functions built into FB 2.5.2, should I try those? If so any
suggestions as to what to do / look for, I have not used the monitoring
tables before.
Thanks
Rob
[Non-text portions of this message have been removed]
Poul Dige
(Sorry for the formatting but Outlook has its own ways...)
Fra: firebird-support@yahoogroups.com [mailto:firebird-support@yahoogroups.com] På vegne af Robert martin
Sendt: 12. december 2012 23:57
Til: firebird-support@yahoogroups.com
Emne: [firebird-support] System performance issue
Hi
One of our clients has a machine where intermittently the system crawls
to almost stop. Launching programs (like paint) takes about 2 minutes
to open. The CPU usage is low (1-3 %) but HDD usage shows at between 97
- 100%. This situation can last 30 mins to 2 hrs before returning to
normal.
I raise the issue here because we have traced the cause to Firebird.
When the machine is on a go slow, stopping the Firebird service fixes
the problem. Sometimes starting the service again and connecting breaks
the system again, sometimes it doesn't. I had though the issue may have
been garbage collection but we have moved from FB 2.0 super server to FB
2.5.2 super classic and are still getting the problem. I'm pretty sure
garbage collection is supposed to be a low priority activity, right?
The system often enters 'slow more' straight after being rebooted.
The weird thing is that while the system is slow and the HDD shows at
around 100% use in resource monitor, the actually data throughput is
about 1 ro 5MB a sec. This is can be much higher when the system runs
normally. This made me think the HDD might be faulty but all
indications are that its not (according to a hardware technician) and of
course the problem fixes itself when stopping Firebird.
System is
i7 quad core (8 cores in Windows)
10 GB Ram
Windows 7 64bit
16GB free hdd space
Firebird 2.5.2 64bit
DB size is 2.7 GB
Any suggestions would be greatly appreciated. I know that there are
monitoring functions built into FB 2.5.2, should I try those? If so any
suggestions as to what to do / look for, I have not used the monitoring
tables before.
Thanks
Rob
[Non-text portions of this message have been removed]