[Tracker-discuss] [pydotorg-www] offline 10-10-18
Izak Burger
izak at upfrontsystems.co.za
Tue Oct 19 09:30:11 CEST 2010
On 18/10/2010 23:29, anatoly techtonik wrote:
> Thanks.
> But I am still curious about what happened and how many hours tracker was down.
> Is there a monitor to mail here if anything went offline?
We have nagios monitoring on that machine. The only problem is, nagios
runs on nina, and nina is the host to the psf virtual host. So sometimes
they go down at the same time, although it usually goes for months
without any problems at all.
Also, it sometimes goes down at night, and I will only notice the next
morning.
But... we need to do something about that machine anyway. We need to
upgrade it, which basically means we need to get a new one, migrate it,
change the ip address, etc etc. Its on my todo list, but I need to do
the costing again and discuss it with Roché.
I can answer the first of your questions immediately. The instability is
not repeatable. The only real thing I did find is that there appears to
be issues with the ext3 file system driver, that will occasionally cause
a corrupted journal and cause segfaults or it will report a full file
system when it isn't. I therefore dropped the journal, as it is in fact
more stable on ext2. To make matters worse, the same thing happened on
my laptop just a few days ago, with a fairly new kernel, so it seems
there is an as yet unreported bug in ext3. But I'm just speculating.
regards,
Izak
More information about the Tracker-discuss
mailing list