Beta Apache Issue

**NexDog** · 05-29-2002, 03:56 AM

Beta is having a rough ride today. Apache has gone haywire and we are having to take the server offline to troubleshoot the problem. Obviously todays recompile of Apache wasn't perfect.

Stay tuned....

nite_messenger · 05-29-2002, 05:10 AM

how long it this go to be?

nite_messenger · 05-29-2002, 05:12 AM

The server status at your support page indicates that the FTP is ok but i encountered "! Socket Error: no DNS entry" when try to connect FTP server.

amit210 · 05-29-2002, 05:18 AM

Well i just hope this is resolved soon.. it's already been about 3 hours..

PaulC · 05-29-2002, 05:32 AM

Whats the problem????

**NexDog** · 05-29-2002, 05:41 AM

Actually, it's more like 2 hours, but it's still bad. Here's an update:

As we know, Apache on Beta failed earlier today and required a complete recompile and reinstall. The only people qualified to do this are the techs at Racksoace. Unfortunately it looks like a mistake was made and now Apache is completely unreachable. We have custom Apache packages on all servers to improve the performance when the network gets busy. Apparently the tech that performed our latest Apache recompile didn't leave any notes on what he did so the late shift at RS are looking at the server and thinking, "What the hell is this?"

They are going to persevere for another half an hour and if they still can't find a solution they are going to call up the tech that did the original config and get him to fix it.

On another note, Sprint is having problems so if you are trying to get to RS from the New York area and Europe, you can't get through either.

These past two days have tested my patience to the limit.

But I would like to add that we haven't experienceed any downtime for about 6 weeks. Little consolation, I know but it was something we were proud of.

**NexDog** · 05-29-2002, 05:48 AM

Paul,

The problem is that we don'tbrun a stanadrd 1024 Apache package. Plesk doubles or triples up on vhosts as you know so once you reach the 400+ mark for domains, Apache is in danger of crashes due to too many file requests. We always configure the Apache kernel to handle around 16000 file requests so it can handle those busy moments.

When Apache crashed this morning it was returning standard 1024 errors which made no sense as that wasn't the config. It locked up and failed to restart under any utility. So it was recompiled with a custom install but wasn't restarted properly to check it's performance. So the server got a little busy and apache gave up. Now the RS techs are having a nice time trying to work out what the last guy did.

It's almost funny. Sorry everyone........

PaulC · 05-29-2002, 06:21 AM

Thats the problem with big companies, nobody knows what other people have done/are doing

nite_messenger · 05-29-2002, 06:27 AM

my site is up now... is the problem solve completely???

**NexDog** · 05-29-2002, 07:00 AM

Well, not completely. Apache is reading a ridiculous amount of vhost entries - almost 2000 even though Plesk registers only 580 or so. Apache will not start with that amount of domains that triggers about 4700 file requests so we've had to remove about 100 vhost entries from Apache's httpd.inlude file until we can find a solution. So a few people will still be affected...........

But we aren't finished yet.