Is this a regular occurence at the moment. As it seems to happen at least once a day for me atm (since monday, and if i remember rightly some of last week, everyday bluurs into the other...).
Browsers. Mozilla 5.0 & IE. 6.0.
May 23 2007 16:09
May 23 2007 18:05
Er, yeah it is unfortunately.
I installed a php cache (it makes our code go faster). But all php caches generate faults sometimes which whites out the screen.
There are workarounds for this - i.e. monitoring the server to see if that's happened, then restarting the cache so it clears the error, but I haven't worked that bit out yet. Or rather I thought I'd worked it out until the same error happened today and showed me I hadn't.
At the same time, for various reasons, not all admins have regular internet access, so despite me teaching everyone how to fix this error manually at the libcom training day, hardly anyone's around to do it. It'll get sorted soon enough.
Sorry 
May 23 2007 19:07
rkn:
white out apache restart
May 23 2007 19:13
it just happened here
May 23 2007 19:17
rkn: tried this but no dice.
if anyone knows about /etc/rc.d or monitoring segmentation faults in debian etch pm me!
May 23 2007 19:55
Mike Harman wrote:
libcom training day.
sorry
May 23 2007 21:02
no joke muthafucka 
May 23 2007 21:21
I doubled apc memory allowance.
If anyone else kicks, please say if it's white screen or not. I don't think we're going down for any other reason at the moment.
worst case, we take apc out (it's only one line in /etc/php5/apache2/php.ini if you need to rkn) and see how the server copes without it. Better to be a bit sluggish than go down every day or so.
May 24 2007 10:26
Joseph K. wrote:
no joke muthafucka
I spiked them all with PCP first too.
Yeah catch is this cos of the new, more expensive server?
May 24 2007 11:13
John no, it's not because of the new server..
It's because I installed an opcode cache which causes apache segmentation faults under certain conditions, and because there's not enough people using opcode caches tfor there to be many howtos or bug reports to steal from. I put one on the old server, and we had one or two of these, but all versions are different on here and there's about 15 or 20 variables which I don't have the time or expertise to go through and debug.
I'm going to switch it off. It might slow things down a bit (please say so if it does), but we shouldn't get that particular fault any more. Then I can switch it on next time I find a workaround.
May 24 2007 11:16
OK I switched it off. Suggest we try without for a couple of weeks or more, see how server load is. Then on for a week if there's a few of us around to kick if necessary, and try to judge how much difference it makes. If it's not much, we can leave it until the server load starts getting high again and try again.
Can comment on articles and discussions
Yeah me too, goes straight to the ‘the page cannot be displayed’ screen, was down for around 45 minutes today, same yesterday. Not afaik at the same time each time.