Firmware version..
- Thread starter marotori
- Start date
If I am correct you are based on the redhat/centos tree?
See...
https://bugzilla.redhat.com/show_bug.cgi?id=790638
I have been fighting this issue for days now.
No amount of tuning eliminates the issue. In short.. High io results in the raid driver going into a hung state.
rob
See...
https://bugzilla.redhat.com/show_bug.cgi?id=790638
I have been fighting this issue for days now.
No amount of tuning eliminates the issue. In short.. High io results in the raid driver going into a hung state.
rob
Have you seen this thread?:
http://forum.proxmox.com/threads/8661-Proxmox-soft-lockups-on-megaraid-controller
The moment I pressed submit I realized you started that thread!
Doh!
http://forum.proxmox.com/threads/8661-Proxmox-soft-lockups-on-megaraid-controller
The moment I pressed submit I realized you started that thread!
Doh!
Last edited:
Sadly.. yes 
I have managed to 'stabilize' the system through various tweaks...
But nothing sorts it once and for all.
To give an idea... just an rsync to migrate a machine will bring the server to it's knees!
It is really quite a major problem!
Consider... pretty much most supermicros.. lsi megaraid. Dell Perc... Lsi megaraid!
This needs to be fixed
I have managed to 'stabilize' the system through various tweaks...
But nothing sorts it once and for all.
To give an idea... just an rsync to migrate a machine will bring the server to it's knees!
It is really quite a major problem!
Consider... pretty much most supermicros.. lsi megaraid. Dell Perc... Lsi megaraid!
This needs to be fixed
I feel your pain
I was an early adopter of Areca 1880 cards that had some firmware bugs and cooling issues (they redesigned the heat sinks to fix that)
To make matters worse, at the same time, I also had over 100 WD RE3 disks that also had firmware bugs.
Between the two we were dealing with locked up servers and failed disks randomly for a few months.
Hope you find a solution soon.
I was an early adopter of Areca 1880 cards that had some firmware bugs and cooling issues (they redesigned the heat sinks to fix that)
To make matters worse, at the same time, I also had over 100 WD RE3 disks that also had firmware bugs.
Between the two we were dealing with locked up servers and failed disks randomly for a few months.
Hope you find a solution soon.
so...
I have gone through every tunable setting possible.. comparing to the previous 1.X series kernels.
Nothing works. Without fail.. high io.. and the system stalls.
What is strange... you would think the disks would be thrashing - but the dont! They are realatively idle, yet the OS is behaving like they are flat out!
My next test is to force an earlier kernel into the system - see if that works!!
Rob
I have gone through every tunable setting possible.. comparing to the previous 1.X series kernels.
Nothing works. Without fail.. high io.. and the system stalls.
What is strange... you would think the disks would be thrashing - but the dont! They are realatively idle, yet the OS is behaving like they are flat out!
My next test is to force an earlier kernel into the system - see if that works!!
Rob
No, because OpenVZ (ioprio) and ionice needs CFQ. But I uploaded a patch
to improve performance on CFQ recently:
https://git.proxmox.com/?p=pve-kernel-2.6.32.git;a=commit;h=c9fd4d7796d4e6c1aa9721c71650011b72ccd97b
I have tuned and tweaked like crazy... Nothing makes the problem go away.
I am now convinced that the issue is sources in a wired driver/card/drive combination.
Changing the drives to a different make helps - but not perfect.
I have sent logs off to lsi. Hopefully they can make some suggestions!
How can I get your patch to test?
Rob
Sent from my iPhone using Tapatalk
I am now convinced that the issue is sources in a wired driver/card/drive combination.
Changing the drives to a different make helps - but not perfect.
I have sent logs off to lsi. Hopefully they can make some suggestions!
How can I get your patch to test?
Rob
Sent from my iPhone using Tapatalk