My box was running fine with 1 GB but it swapped a lot. So yesterday i ordered an extra 1 GB, just a couple of minutes after they upgraded the box i logged into plesk and was very happy to see 2 GB of ram now. Since then its was a nightmare. My box crashed right away. Until today i could not log into SSH, Mail, ftp etc. Everything was ultra slow, a webpage or plesk takes 15 minutes to load it its load. Now after hours and hours im logged, the box doesnt swap anymore, it doesnt use all memory, but still, you cant do anything with it, it works slow, slow, not mail, websites takes forever, plesk forget it. It was running perfect with 1 GB of Ram and now with 2 GB the box is slow. How come that hardware make the box run worst then it was. It worked very fine with 1 GB, and it only had 20 or 30 minutes peak time of website with forum, then it was froozen and slow, but only 30 minutes a day. Since i have the 2 GB of ram just a couple of hours it runs slow all the time if it works. Before it worked at least 23 hours a day, now its almos 24 hours and i cant use it.

The iowait is also very high, it was always a little high but not like now. Could it be that TP put a bad stick of mem on the box or what? I think i will go back just to the 1 GB. The other thing is maybe that now that i have 2 GB of ram i should tune better again the mysql config, but i really dont thing thats the problem, if it worked with 1 GB why not with 2 GB.
Im lost here. I put for example on the SSH
shutdown -r now

And it doesnt reboot, now matter how many time i give it the command. It doesnt respond to almost anything only apache seems to work some how since websites load.

It is very possible that it was bad memory, the fact that things are actively crashing would be a red flag. Have them replace the memory ASAP.

Well the iowait load is all the time on 98% now. They told me apache was crashing the server, how funny it was working very well with the 1 GB stick.

Its very funny what you said about to replace the memory ASAP.

They take 8 hours only to make a memory test, and that is after calling by phone. Its going to be 24 hours that my server is down. Support is slowwwwwwwwwwwwwwwwwwww.

But the clients i have on the webhost are already threating me with going to another host.

I WILL NEVER AGAIN BUY AN UPGRADE FROM THE PLANET. I ordered a stick of RAM that costs 25$ a month and this is what i get my server down. Not to mention they charge 25$ setup to put a stick of memory. At least they can test the stick before sticking it on a server. I dont know what the hell is the problem except my server crashes 5 minutes after rebooting, all i know is that the problem started 20 minutes after they put the stick on the server. I cannot believe how slow support is. Imagine 24 hours without email. Does it take so long to replace the stick and test if thats the problem?

A thorough memory test can take a minimum of 8 hours, however if I was a support technician, I would replace first and test on the bench at my leisure. You're dealing with fallible, probably overworked humans, try making that suggestion and see where it leads? Keep us updated on your progress

8 hours to test 1 GB of Ram?Wow. Well ist it easier to put in the meantime another stick and test the ram in the machine of the tech support department instead mine which is live.
Well at least they could put or keep updated the ticket, i dont anything, just that the server is down.
I just received another call from another client that will cancel his email with me because of the downtime. Thanks a lot The Planet !!! I can really put my business on your hands.

> 8 hours to test 1 GB of Ram?Wow. Well ist it easier to put in the meantime another stick and test the ram in the machine of the tech support department instead mine which is live.

Yes, that's exactly what I said above. Insist they replace it now and test it later.

Going though the exact same thing right now-
Server was running @ 0.89 - 1.3 (game server)
Added another 2 gig of mem, and server shoots up to 4.00-9.00%

-----
So I submit support ticket, they want 4 hrs to test the mem-

My response was pull it out test it on somebody elses time.
I gave them 3 options,
pull mem put in new stick
Pull mem, test it, then put it back

or pull the entire server, when I move my clients to another box that I have.
(65.00 to put in 2 gig of mem) is way beyond unreal anways..

looking forward to their response, their lose on 500+ revenue on a server vs a 65.00 mem stick.
But I'm not paying for the downtime for something that should be working to begin with.
And to even suggest that a game server be down for 4 hrs with x amount of clients-- that's funny to begin with.

(you sales/marketing types) want to watch the lost revenue of a 500.00 server on -- 1560547PLNT
on some BS testing methodolody.. Pull the damn memory, test it on your own dime--

Well My box will be exactly in 1 hour be 24 horus down. All for the damm Ram stick. I had to finish installing things on my software and i cant since yerterday. The whole machine is down, im freaking up now, im starting to look hosts alternatives to change to another server, it will take me less to change server then to be down for whole day, i cant lose any client more, i really cant afford to be offline 24 hours, i will lose all the client that toke me a year to get, no to mention the time i spend to market my website. Im sure the planet will not pay my clients for the offline time. How is it that i buy addons for then and i get downtime instead. If this isnt fixed by tomorrow at 5 i will start moving my DNS. I dont care to pay more for another server on another company, that will be steal best then lose all my clients thanks to that fu*** ram stick.

I got back from techs, its not hardware, they tested all. Box went online again for 2 hours. Now the same problem again. Again iowait cpu is at 98%, it was at 0% when box was working again. Suddenly again the problem raises and iowait cpu usage goes to hell. How do i find out what is causing this?

4224 root 25 0 620 620 264 R 64.7 0.0 1:15 0 gzip
4254 root 25 0 1272 1272 616 S 11.3 0.0 0:12 0 mysqldump
2568 mysql 15 0 27216 26M 3148 S 7.9 1.3 0:08 0 mysqld



Thats causing the load after all the services are shut down.

I think i may have found the cause. I rebooted the server and turned all service off, apache, mail, psa etc.

I have this:
3069 root 25 0 612 612 264 R 70.7 0.0 0:31 0 gzip
3135 root 25 0 1056 1056 616 S 5.5 0.0 0:04 0 mysqldump

Its mainly a gzip.


I dont know if this makes sense but i think here is the problem.

I programmed the RAM upgrade to be exactly at the same time where plesks makes backup of 3 websites of me. The iowait is high when it makes backups and i think plesk uses gzip to compress the backups wich then are send to my NAS FTP.

Since the upgrade was at the same time i think it broke something and plesk is trying to make all the time the backup now, since its daily.

I just dont know how to make it stop, it seems it starts again after every server boot.

You can try pkill gzip
and that should kill that process

It comes and goes. Now for example it did not come anymore in the last 10 minutes. Wehn it comes it takes all the iowait. I will try to kill it, i did before but it seems to restart itself on the reboot. And what it had to do with the ram upgrade i still dont know. Somehow it wonders me too how much apache takes now that the other gzip is not:
CPU
5783 apache 15 0 27160 26M 8508 S 5.5 1.3 0:05 0 httpd
4672 apache 15 0 28612 27M 8512 S 4.7 1.3 0:09 0 httpd
4789 apache 15 0 28252 27M 8784 S 4.5 1.3 0:10 0 httpd

How can i see which process or monitor the apache, maybe someone is exploting the server somehow and via apache and using it for spam. Maybe thats why together with the gzip it freezes all.

And here comes again the gzip:

PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME CPU COMMAND
7020 root 25 0 616 616 264 R 63.1 0.0 0:54 0 gzip
7019 root 25 0 10776 10M 2084 R 5.9 0.5 0:05 0 PleskX.pl
7052 root 17 0 11000 10M 2084 S 5.5 0.5 0:03 0 PleskX.pl
7053 root 20 0 1084 1084 616 S 4.5 0.0 0:06 0 mysqldump
7010 mysql 15 0 46892 45M 3296 S 2.9 2.2 0:05 0 mysqld
7021 root 25 0 10876 10M 2068 S 1.9 0.5 0:01 0 PleskX.pl

I know its dumb what i do, but i dont have anything else but to sit for 1 hour and monitor the box, since im sick now. I sleep 1 hour in 2 days, finally yersterday when the box was running i went to bed until to be wakeup again tomorrow, and surprice the mails again down. I restarted the server now, i will monitor it until it freezes agian.

I uploaded my.cnf for mysql again and it seems to be working now, lets hope it stays so. Could it be that the my.cnf ist not well enought now that the system has 2 GB of RAm and is causing a bottle neck on the hard drive?

This one is running, i used this for 1 GB not sure if its the same for 2 GB now, maybe it could be a bottle neck.