1

I have multiple websites hosted on a Linode VPS and I'm having an issue with one of them: every page that I try to load has about 10 seconds latency.

Apache logs are clean and the other websites on the same machine are running well.

At a first glance I tought it was a memory problem since the VPS has got only 512M, but from the linode dashboard CPU and Disk I/O are normal.

Anyway here we have the ram status:

$ free -m
         total       used       free     shared    buffers     cached
Mem:           487        463         23          0          2         55
-/+ buffers/cache:        404         82
Swap:          255        155        100

Only 23M free, but if it was a memory problem why other websites are going as usual?

I took a live capture with wireshark, and there are some duplicates SYN ACK packets just before the 10 seconds gap.

I'm out of ideas, looking for some clues.

Wireshark live capture screenshot

As you can see from the image, the gap is after the last bad tcp.

Thank you in advance.

UPDATE

I've checked Apache2 logs in debug error level, and this is where something is appening:

151.97.156.191 - - [14/Nov/2012:11:19:40 +0100] [www.fmaisi.it/sid#7f32c625a220][rid#7f32c6801578/subreq] (3) [perdir /home/fmaisi/sites/www.fmaisi.it/public_html/] applying pattern '^index\.php$' to uri 'index.php'
151.97.156.191 - - [14/Nov/2012:11:19:40 +0100] [www.fmaisi.it/sid#7f32c625a220][rid#7f32c6801578/subreq] (1) [perdir /home/fmaisi/sites/www.fmaisi.it/public_html/] pass through /home/fmaisi/sites/www.fmaisi.it/public_html/index.php
151.97.156.191 - - [14/Nov/2012:11:19:54 +0100] [www.fmaisi.it/sid#7f32c625a220][rid#7f32c6537c78/initial] (3) [perdir /home/fmaisi/sites/www.fmaisi.it/public_html/] strip per-dir prefix: /home/fmaisi/sites/www.fmaisi.it/public_html/wp-content/plugins/wp-filebase/wp-filebase_css.php -> wp-content/plugins/wp-filebase/wp-filebase_css.php
151.97.156.191 - - [14/Nov/2012:11:19:54 +0100] [www.fmaisi.it/sid#7f32c625a220][rid#7f32c6537c78/initial] (3) [perdir /home/fmaisi/sites/www.fmaisi.it/public_html/] applying pattern '^index\.php$' to uri 'wp-content/plugins/wp-filebase/wp-filebase_css.php'

As you can see there is a gap of 14 seconds after the pass through of index.php. Any suggestions? I'm out of ideas again.

  • Please note that you potentially have 23+55=78M of free memory: http://www.linuxatemyram.com – Ketan Maheshwari Nov 12 '12 at 20:02
  • There's no evidence of a memory issue. 23MB free is perfectly normal. Is there some reason you think you'd need more than that? Or some reason you think that's unusually high? You only need free memory to account for memory allocations in very unusual contexts, such as interrupt handlers, where memory can't be directly shifted from one use to another. 23MB seems like it would be enough for any such urgent needs. – David Schwartz Nov 12 '12 at 22:03
  • @DavidSchwartz Thank you for the clarification, I was only guessing and I thought that 23M weren't enough. Now I think that I have to inspect what apache2 is actually doing in those 10 seconds – Mistero Lupo Nov 13 '12 at 10:09

0 Answers0