@Sven Same happened to me today for the very first time. Would you mind elaborating just a little bit? I know a thing or two about computers, but I'm no hardcore techy. When you say this error message originates from the tcp/ip stack, then what exactly does that mean? Are we talking performance related issues in general (too many connections, too much bandwidth usage, ..) or is it really hard to narrow it down?
I ran some extensive tests over the last 2 days and I'm 99% sure that steelbone is correct and this error can be traced back to GSA SEO Indexer.
Whenever I run it on low (50-100) threads, I don't ever get this error message. But as soon as I crank it up to high thread counts (500-800) it's only a matter of minutes/hours until the 'Buffer Space' errors pop up.
I figured you might be interest in this Sven, since you stated that it's highly unlikely that SEO indexer causes this.
@Sven A week ago I started to get that error: "No buffer space available". I haven't changed anything, Ser is always up to date (v8.97), and I did all Windows Server updates. Ser is running at 100 threads and with CB, nothing else running in this VPS. Usually it starts after a day, then i restart Ser and it works again for a day.
I contacted my VPS provider PowerUpHosting, and their senior system administrator replied:
We don't have any kind of restrictions on our end, I would advise you to contact the developer and ask him what exactly the issue. It could be possible that one of the recent updates on GSA might have started causing this issue again.
What would you suggest, Sven? Thanks.
Edit: I am trying with scheduler to run half of the active projects at a time. Maybe this helps...
such an error is indeed caused by a memory leak. However I never had one recently using my profilers. I will test again. Since what version did you get this?
@Sven It started about a week ago (maybe v8.94 - 8.95 ?).. I think I know now why... Until a week ago I had fresh lists and the daily limits got done pretty fast, keeping the Memory consumption around 1.00GB. During the week the projects have been running 24/7, raising the Memory consumption up to 1.60GB. And then is when the error comes. Yesterday I started scheduler (keeping half of projects active), the Memory consumption is 800MB and there is no error yet.
Anyone has a solution to this problem? I have those errors poping up all the time for like a month and NOONE knows a solution to it, I have contacted the guy who sold me the email accounts, my VPS provider, posted like 3 million threads here in the forum, not even Sven knows the solution. WTF
@Sven after updating from 9.91 to 9.93 getting this error again: POP3 Login failed - SockError: No Buffer space available (didn't get this error since i wrote here months ago)
maybe some new update might be causing memory leak? no?
Comments
Would you mind elaborating just a little bit? I know a thing or two about computers, but I'm no hardcore techy. When you say this error message originates from the tcp/ip stack, then what exactly does that mean?
Are we talking performance related issues in general (too many connections, too much bandwidth usage, ..) or is it really hard to narrow it down?
@sven
I ran some extensive tests over the last 2 days and I'm 99% sure that steelbone is correct and this error can be traced back to GSA SEO Indexer.
Whenever I run it on low (50-100) threads, I don't ever get this error message. But as soon as I crank it up to high thread counts (500-800) it's only a matter of minutes/hours until the 'Buffer Space' errors pop up.
I figured you might be interest in this Sven, since you stated that it's highly unlikely that SEO indexer causes this.
And still this would not be a problem of the software but your network/OS. Some systems have a certain limit of how many connections they accept.
Read here: http://blog.yaronmaor.net/2010/03/fix-windows-xpvista-connection-limit.html
POP3 Login failed - SockError: No Buffer space available
(didn't get this error since i wrote here months ago)
maybe some new update might be causing memory leak? no?
Actually it happens when or proxies go bad for some time.