|
Hi all,
I am trying to run rendered daemon and when I try to render multiple tiles at a time, it is showing error mentioned in the figure. I am using following configuration to run the Tileserver for whole world Map: Rendering Server: 16GB RAM with 2 CPUs Database Server: 16GB RAM with 2 CPUs I am not sure what is this error about. Could somebody help me to figure out the problem? I pre-rendered tiles till zoom level 14. Thats why I thought of using servers with less number of CPUs and moderate memory. Please let me know if you need any more details regarding the configuration or setup. Thanks in Advance Subin |
|
This appears to be essentially a duplicate of this question. You're trying to use a cloud service for something it's not really designed for. |
Can you post a bit more of the relevant log somewhere (as text, not just a picture)? Of particular interest would be what messages appear when you start or restart renderd.
Also - you say you have problems with rendering multiple tiles; can you render single tiles OK?
Thanks for your reply.
The following log details are the only relevant logs available from Rendering server.
DEBUG: Got Incoming connection, fd 12, number 1
DEBUG: Got incoming request with protocol version 2
DEBUG: Got command RenderLow fd(12) xml(ajt), z(18), x(137843), y(90010), mime(image/png), options()
DEBUG: Sending NotDone response(4)
DEBUG: Sending render cmd(1 ajt 18/137843/90010) with prpotocol version 2 to fd 12
WARNING: Failed to send render cmd on fd 12
In the PostgreSQL server (using AWS RDS instance), I could see following logs:
LOG: Could not send data to client: connection timed out.
LOG: FATAL: Connection to client lost.
As I mentioned in my question, I am using database server with 16GB of memory with 2 CPUs.
Are those above logs occurred because of SQL query is taking too much time to complete and thus rendering server timedout the particular connection?
Thanks in Advance Subin
Can you give us a few more clues? I notice "AWS RDS" is mentioned there, but I frankly don't believe that these are the only logs available.
What I'm guessing is happening is that the disk on which the postgres database is stored is far far too slow. This was the answer given at https://help.openstreetmap.org/questions/62113/not-getting-output-from-tileserver-when-i-reduce-the-aws-rds-memory-size-to-16gb/62132 and yet you still seem to be persisting with this configuration.
I'd suggest that you follow the advice that you've already been given ("For a world-wide database you must have your database on SSD, and ideally the database should be local to the rendering machine and not connected to over the network.").