H2 DB is reopened on every request

[XANi]

I've tried default install (just installing package, no setup) and I've noticed it is awfully slow, even with just one job in the system. Strace after every request looks a bit like that (that was just clicking nodes buttons, or jobs button):

[pid 26787] 11:48:56 open("/var/lib/rundeck/data/rundeckdb.h2.db", O_RDWR|O_CREAT, 0666) = 211
[pid 26787] 11:48:56 fstat(211, {st_mode=S_IFREG|0664, st_size=163840, ...}) = 0

and then it gets closed once request ends:

[pid 26787] 11:48:57 lseek(211, 0, SEEK_CUR) = 8192
[pid 26787] 11:48:57 lseek(211, 0, SEEK_END) = 2260992
[pid 26787] 11:48:57 lseek(211, 8192, SEEK_SET) = 8192
[pid 26787] 11:48:57 lseek(211, 0, SEEK_CUR) = 8192
[pid 26787] 11:48:57 ftruncate(211, 163840) = 0
[pid 26787] 11:48:57 lseek(211, 8192, SEEK_SET) = 8192
[pid 26787] 11:48:57 close(211)         = 0

Shouldn't it be kept open ? It wastes a lot of time first trying to lock db, read headers than read data and that happens on each request

[ahonor]

@XANi what are your JVM memory settings (i.e. max heap size)?

[XANi]

Just the default package uses (rundeck-2.6.2-1-GA.deb) which is -Xmx1024m -Xms256m -XX:MaxPermSize=256m

I don't think it is gc/memory related issue, here is snapshot from jstat when running

☠ jstat -gc 23175 5000
 S0C    S1C    S0U    S1U      EC       EU        OC         OU       PC     PU    YGC     YGCT    FGC    FGCT     GCT
41472.0 43008.0  0.0    0.0   254464.0 70889.5   290304.0   157304.3  175616.0 88122.6     15    0.662   1      1.281    1.943
41472.0 43008.0  0.0    0.0   254464.0 70889.5   290304.0   157304.3  175616.0 88122.6     15    0.662   1      1.281    1.943
41472.0 43008.0  0.0    0.0   254464.0 75150.9   290304.0   157304.3  175616.0 88163.2     15    0.662   1      1.281    1.943
41472.0 43008.0  0.0    0.0   254464.0 89202.1   290304.0   157304.3  175616.0 88281.7     15    0.662   1      1.281    1.943
41472.0 43008.0  0.0    0.0   254464.0 124674.7  290304.0   157304.3  175616.0 88649.4     15    0.662   1      1.281    1.943
41472.0 43008.0  0.0    0.0   254464.0 127228.4  290304.0   157304.3  175616.0 88664.9     15    0.662   1      1.281    1.943
41472.0 43008.0  0.0    0.0   254464.0 151224.9  290304.0   157304.3  175616.0 89390.6     15    0.662   1      1.281    1.943
41472.0 43008.0  0.0    0.0   254464.0 151224.9  290304.0   157304.3  175616.0 89390.6     15    0.662   1      1.281    1.943
41472.0 43008.0  0.0    0.0   254464.0 178098.6  290304.0   157304.3  175616.0 89440.5     15    0.662   1      1.281    1.943
41472.0 43008.0  0.0    0.0   254464.0 191932.5  290304.0   157304.3  175616.0 89462.3     15    0.662   1      1.281    1.943
41472.0 43008.0  0.0    0.0   254464.0 194051.4  290304.0   157304.3  175616.0 89507.4     15    0.662   1      1.281    1.943
41472.0 43008.0  0.0    0.0   254464.0 204478.3  290304.0   157304.3  175616.0 89531.2     15    0.662   1      1.281    1.943
41472.0 43008.0  0.0    0.0   254464.0 211992.1  290304.0   157304.3  175616.0 89571.7     15    0.662   1      1.281    1.943
41472.0 43008.0  0.0    0.0   254464.0 217758.9  290304.0   157304.3  175616.0 89576.6     15    0.662   1      1.281    1.943

Looks like GC wasnt even ran during those requests

I was just clicking jobs and nodes tabs few times (totally empty clean install, 1 project with zero tasks) and all requests took more than 6 seconds (normal hard drive, no ssd)

[ahonor]

@XANi what is the Model Source for the project?

[XANi]

File. It is just default install, I haven't changed anything

[XANi]

I've tried to move whole /var/lib onto ramdisk/tmpfs and it made requests take ~1s or shorter so it is definitely something related to disk access

[ahonor]

Please reopen if this remains a concern

[XANi]

It looks faster in 2.10 but I can't really compare because after upgrading to 2.10 all jobs disappeared... i just had them in yaml dir but now that isn't even an option and I'm confused why...

[ahonor]

@XANi Rundeck does not automatically load job definitions from a directory. Try using the rd CLI to reload them.