Ran out of disk space


#1

Hello,

This post is mainly to provide documentation of what happens when you run out of disk space. I tried to load a file containing 100 vectors each 42315 in length, of type double and I ran out disk space. Of course the problem appears to be solved now that I’ve increased the disk space!

scidb@ubuntu:~$ iquery -naq "load (eigVect, '/home/scidb/export.txt')"
SystemException in file: src/network/BaseConnection.cpp function: sendAndReadMessage line: 367
Error id: scidb::SCIDB_SE_NETWORK::SCIDB_LE_CANT_SEND_RECEIVE
Error description: Network error. Cannot send or receive network messages.

from a separate, previously open client instance I tried to see if anything had loaded and got a similar error and the client stopped:

AFL% between(eigVect,0,0,20,0);
SystemException in file: src/network/BaseConnection.cpp function: sendAndReadMessage line: 367
Error id: scidb::SCIDB_SE_NETWORK::SCIDB_LE_CANT_SEND_RECEIVE
Error description: Network error. Cannot send or receive network messages.

I tried to restart scidb and got this error message:

8 [0x7f7fc0f047a0] FATAL scidb.services.network null - Error #system:111 when connecting to localhost:1239
8 [0x7f7fc0f047a0] FATAL scidb.services.network null - Error #system:111 when connecting to localhost:1239
iquery SystemException in file: src/network/BaseConnection.cpp function: connect line: 267
Error id: scidb::SCIDB_SE_NETWORK::SCIDB_LE_CONNECTION_ERROR
Error description: Network error. Error #system:111 when connecting to localhost:1239.

Cheers,
Dave

ps. I’m not sure why the [code] blocks above have the phrase “SELECT ALL” associated with them…


#2

Hi diahr …

Some details. When you ask scidb --version, what version of SciDB are you on? 12.3? 11.12?

Second - can you post the show( eigVect ), please? I’d like to see the contents of the target for your load, and the first line or two of your export.txt, because I that will help me clarify something. I’m worried because loading 100 vectors of 42315 doubles is only about 32 Meg. I’m very surprised that you’re running out of disk space. . . even before you increased it.

To your broader point.

  1. What happens at the moment is that the iquery tool connects to a running SciDB instance. At the moment we focus all iquery connections to a single instance, for no real reason other than it’s convenient for most users who want to log / follow what’s going on. (We plan to relax this in the future as we get users with more concurrent connections.) When the running SciBD instance gets a trap (in this case, a write failure I would presume: check scidb.log or scidb-stderr.log to confirm) we take the precaution of killing the process and restarting it. At that point, your other iquery client finds it’s connected to a dead process, and will report the FATAL scidb.services.network null - Error #system:111 when connecting to localhost:1239 error when you ran the between(eigVect,0,0,20,0) query.

  2. Your “tried to restart scidb and got this error message” worries me a lot more. Did you just go scidb.py startall <your_installation>? Or did you use stopall? There should be no need to start the server again after this kind of problem. We have the restart automatically. What commands did you use, here?

Paul


#3

Ahh, I think we’ve addressed everything over on this thread:
viewtopic.php?f=11&t=598

Thanks,
Dave