Portal Home > Knowledgebase > Articles Database > Acunett horrible experience, need suggestion on other management
Acunett horrible experience, need suggestion on other management
| Posted by whdev, 11-01-2009, 09:22 PM |
| Recently we are having problems with our server. We're running nginx and nfs on a server, and load would go really high with nothing but JUST nfs running. NFS shouldn't cause high load and it's unusual. If we run nginx as well, the load would go even higher and service would fail periodically (we have a port 80 monitoring script running elsewhere to monitor this). Wasn't like that previously when everything is ok. The problem usually occurs at peak time so there's only a small window to investigate.
We have acunett as our server management company and we informed them a couple times of the issue. BUT all they do is monitor the load, wait till it's after peak hours, and tell us everything is normal. A couple times, and they just do the same damn thing!
It's a frustrating experience for us and not being able to fix this issue for nearly 1 month already.
Anybody has any other good server management people to recommend for some specific issue like this?
Thank you.
|
| Posted by Spudstr, 11-01-2009, 11:23 PM |
| several factors including high load. the biggest one popping to mind is your disk speed and how "busy" that nfs spindles are. Common problem is slow set of drives, the nfs clients read a lot faster than they can distribute to the end user from the nfs server, hence they spin their wheels on busy busy servers.
|
| Posted by TheServerExperts, 11-01-2009, 11:26 PM |
| Sorry to hear about your bad experience, i know acunett has a representative on WHT lets hope he can come in and chime why they didn't fix your problem.
|
| Posted by AcuNett, 11-03-2009, 01:17 AM |
| You mentioned there is a small window to investigate, and more often that not that small window is not going to allow us to track the source of the problem immediately. However, no one should have told you the issue is fine. If you could let me know what the ticket ID is on the issue, I can look into this for you.
|
| Posted by AcuNett, 11-03-2009, 01:23 AM |
| I looked up your ticket, which was filed just yesterday, and I must say I'm disappointed to see such a wild exaggeration in this post.
We told you it was normal one time. We were not informed this was a recurring issue nor that it happened at a specific time. You stated that the load was high and that it never happened before.
Thus, we monitored the server closely for several hours, found the load was normal, nothing abnormal was running, and reported back to you. Load averages can rise due to any number of reasons, and temporary spikes (and your reported load of 3.0 is by no means a huge spike) are completely normal on a web server.
I'm sad to see your unhappiness with our service. I don't like to see an unsatisfied client, be it an issue on our end or not. Please send in a ticket and we'll gladly refund your money.
Last edited by AcuNett; 11-03-2009 at 01:31 AM.
|
| Posted by steveks, 11-03-2009, 09:43 AM |
| Why can't your tech write a bash script & add it to cron which can track the processes running on the server?
I am just seeing an ordinary work from your techs. All the issues can't be solve simply by having a glance. Some need research & investigation.
Glad, you are able to give them refund.
|
| Posted by activelobby4u, 11-03-2009, 11:03 AM |
| The solution is to catch the load increase in early stages by setting up alerts using a shell script or monitoring system like nagios.
However when the load increases beyond a 25 or 30 mark (when monitored using a shell script), the email has no guarantees to reach your mailbox as the queue might get frozen due to the load
|
| Posted by AcuNett, 11-03-2009, 01:49 PM |
| That is not sensible when asked to check a temporary load issue. If the load is temporary you go in, and you find out what's causing it, hopefully before that spike is over. Writing a script is a complete waste of time, when by the time your done the load issue will be gone. Let me re-iterate once more that the customer wanted to know right then and there what was causing the load at that instant. He was not asking us to track a recurring load problem.
Thank you for your suggestions, however your proposals are applicable only to recurring load problems.
Sandy, I'm not sure where email plays a factor in this, and as mentioned before the load never went about 3.
|
| Posted by activelobby4u, 11-03-2009, 01:54 PM |
| Thats amusing . When the Op said "load would go really high with nothing but JUST nfs running." , I was expecting somewhere above 25 . If 3 is what he proposes as high load, I would definitely stand with you on this.
|
| Posted by UNIXy, 11-03-2009, 02:10 PM |
| It's obvious that NFS or the internal network is the bottleneck here. Can you go over the setup/configuration?
Regards
Joe
|
| Posted by AcuNett, 11-03-2009, 02:18 PM |
| Since he responded back yesterday letting us know it is a recurring issue, we were able to track it with more depth
The source of the problem lies with a rogue NFS client. He's receiving 1,000 packets per second to his server once NFS is started. Hard disk input rates increases six-fold and load increases from 1.0 to 3.0 during this time.
We're waiting for his response before tracking the rogue client.
|
| Posted by whdev, 11-03-2009, 10:01 PM |
| There isn't just ONE ticket. There's multiple. Ticket TNI-786443 is one, ticket 138052 is the other one. Problem first reported: 06 Oct 2009 10:05 AM.
Both stated in ticket 138052.
3 is perfectly normal for webservers, but that's why I asked one of the techs to look beyond the server load (which Ronny did and found the problem with the packets). I even pasted the vmstat in one of the ticket and the tech said there's nothing wrong with it. The server was having server load < 1 previously, and suddenly shot > 7 at at times for close to so many days is an indication that something is wrong. I'm getting loads of timeouts on the web server front. That isn't normal. Server load is just an indication.
All being said, I believe in Ronny's capability and I'm relieved Ronny is now working on the issue personally. I'll keep everyone updated with the progress.
Last edited by whdev; 11-03-2009 at 10:09 PM.
|
| Posted by AcuNett, 11-03-2009, 11:04 PM |
| TNI-786443 is the ticket I've been referring to which was opened November 1st, 2009. I have no idea about 138052. That ticket is not in our system. I would need the first three letters as no tickets filed under your account end with the numbers "138052".
|
| Posted by steveks, 11-03-2009, 11:36 PM |
| As a Sys-administrator, the bash script could have helped much more on this case. But, I was pretty happy atleast Ronny was able to find your issue.
But, its silly that a person needs to come to WHT to solve his issue. lol
|
| Posted by IPswing-Sarwar, 11-09-2009, 09:01 PM |
| I have had horrible experience with them too... what is more annoying then there response time (which takes days) is that this guy never admits anything...
take my words, go with touchsupport and let them handle your server, that's what i did.
|
| Posted by Prolime Servers, 11-11-2009, 06:44 PM |
| we are with acunett from long and they manage many servers for us
they are no doubt very good company thought i agree on a small point that there tech just say that we are monitoring and loads automatically comes down
but regarding response time which you are mentioning is totally wrong
they always reply with in 5-10 minutes. they are no doubt top of the notch company.
best of luck acunett
|
| Posted by jdk, 11-24-2009, 09:35 PM |
| I don't know about the 5-10 minute response times, but I too have been with Acunett for several years and I have never had any issues with their support. They are top notch and Ronny and his team are very knowledgeable. There has never been an issue they could not resolve.
|
Add to Favourites 
Print this Article
Also Read
