Hello,

So we have a setup where we're running a bunch of servers from which we generate a map that gets displayed on our website.

The setup uses a 'master' server which runs mapnick, renderd, an apache 2 server and a postgres database, plus at least 2 slave servers that only run the mapnick,renderd and apache2 server and we have them using the postgres database found on the master server. All servers running ubuntu, the slaves have a 4 cpu cores and 8 gb of ram.

Lately we've been seeing the slave servers using up all the memory, the unix kernel kills the renderd process (which is the one using the most memory). And then we have a monit process that restarts it, so the servers keep going. Tried to change the number of threads used in rederd.conf down to only 1 thread, but that just made the memory usage go down slower.

This is the graph we see on our servers: https://dl.dropboxusercontent.com/content_link/6GHw57e25indiTnh5O3WuSM6ITMKGdbLq8iW0mvjeT4UxbQraMolPkTRbJc8meXT/file

The one on the left is using 1 thread, and the one on the right 2 threads.

Any ideas what might be causing this? From the graphs we see that the memory usage suddenly spikes, can this be caused by some cached tiles expiring or something similar to that?

Kind regards,

Tiberiu Ionescu

asked 10 Jun '16, 14:03

Tiberiu22's gravatar image

Tiberiu22
56115
accept rate: 0%

edited 13 Jun '16, 08:27

1

meta: could you please link to an image instead of a html page with dozens of necessary (third-party) scripts? Best would be to upload here. If you do not have enough rights for this yet, just give us the permission to re-upload here, then someone can replace the externally hosted image.

(10 Jun '16, 19:45) aseerel4c26 ♦

Hello,

Did you find any solution since then ? I am having the same issue:

  • renderd works fine and renders about a dozen of tiles
  • suddenly the unix kernel kills the renderd process because of excessive memory usage

I tried using a very small database and reduced the number of threads to 1 in renderd.conf but the problem is still there.

Any idea where this might come from ?

permanent link

answered 03 Sep '18, 13:20

sammy's gravatar image

sammy
161
accept rate: 0%

I would have suggested reducing "num_threads" in your "renderd.conf". However you've already done that...

I'd have a look at what's using the memory when it runs out. If 1 renderd process is using 8Gb, then I wonder if there's a leak somewhere - maybe check versions of things to see if something needs updating?

permanent link

answered 10 Jun '16, 17:26

SomeoneElse's gravatar image

SomeoneElse ♦
32.3k63335756
accept rate: 15%

edited 10 Jun '16, 17:29

Hey! Thanks for the suggestions. Any suggestions on how I can do this? Can I see the operations that renderd is doing and how much memory they're using?

(13 Jun '16, 10:24) Tiberiu22

I reported similar problem here: https://github.com/openstreetmap/mod_tile/issues/181 and here https://help.openstreetmap.org/questions/70490/renderd-out-of-memory. Unfortunately without any feedback...

permanent link

answered 16 Sep, 08:54

stevo_001's gravatar image

stevo_001
112
accept rate: 0%

Your answer
toggle preview

Follow this question

By Email:

Once you sign in you will be able to subscribe for any updates here

By RSS:

Answers

Answers and Comments

Markdown Basics

  • *italic* or _italic_
  • **bold** or __bold__
  • link:[text](http://url.com/ "title")
  • image?![alt text](/path/img.jpg "title")
  • numbered list: 1. Foo 2. Bar
  • to add a line break simply add two spaces to where you would like the new line to be.
  • basic HTML tags are also supported

Question tags:

×85
×14
×8

question asked: 10 Jun '16, 14:03

question was seen: 2,549 times

last updated: 16 Sep, 08:54

powered by OSQA