Stream stopping and not showing old data Seedlink Server as datasource

[himmetgencer]

Hi,

I am using swarm 3.1.0 and 3.2.0-snapshot with a seedlink server as datasource. After some hours/minutes stream stop and old data is not loaded. Also when the user clicks in the empty part of helicorder, the zoom windows shows data.

I tested it with option -Xmx1024m, but this issue continues. Also I'm using SeisGram2K and jAmaseis software, everything is correct.

OS: Windows-10 64-Bit Swarm version: 3.1.0 - 3.2.0-snapshot

Regards...

[dlnorgaard]

About how many hours will it run before it stops updating? I am trying to reproduce this.

[himmetgencer]

Hi, thanks for answer,

Sometimes 4-5 hour, sometimes after 12-13 hours. it's actually random. Also old data is not loaded.

Also you can find screenshot of jAmaseis software.

[dlnorgaard]

I haven't been able to reproduce the problem with the stream stopping, but I made some changes that may or may not help. Can you try the attached? I also made some improvements for getting past data, though I need to work on it some more.

[himmetgencer]

Thank you, I will try it and I will send results.

Regards...

[himmetgencer]

Hi again,

I've tried version 3.2.0 . But stream stopped after 30 minutes. But jAmaseis software continues collect data. Also past data is not loaded.

Regards...

Windows 10 64-Bit Swarm 3.2.0 200 sps x 3 axes accelerometer (STEIM1)

[himmetgencer]

Also, I'm uploading another screenshot of swarm console.

[dlnorgaard]

Well, that's not good that it got worse :( Can you see how much heap space is allocated? Also what is your helicorder setting? I don't suppose your SeedLink server is accessible to me to troubleshoot? With the servers I have available I'm able to run helicorders for days without any stream break.

[himmetgencer]

Thanks for answer,

I sent seedlink server information to "dnorgaard@usgs.gov". I will keep it open for 48 hours because of company policy.

I'm using default helicorder settings. Only I run the software with "-Xmx1024m" option.

I'll send heap size is allocated.

Regards...

[himmetgencer]

You can see heap size in below screenshot. Have you received information of seedlink server?

Regards...

[dlnorgaard]

This version has some improvements, but does not sufficiently address the delays in processing SeedLink data with high sample rates.

swarm-3.2.0-snapshot-bin.zip

[dlnorgaard]

It appears the caching mechanism of Swarm slows things down quite a bit the more data is stored in cache for a helicorder. The high sample rate compounds the problem. Over time, what appears to be a stopped helicorder is just processing data really slowly. Right now SeedLink data source in Swarm does not support no-caching option. The best bet in the short term is to to consider alternative features in Swarm, such as the real-time monitoring, or use helicorder with shorter time period (e.g. six hours) since that reduces the data required in cache.

[himmetgencer]

Thanks for answer, I will try short term alternative features next week. But this issue needs to be resolved.

Regards...

[dlnorgaard]

My understanding is that SeedLink is a real-time data acquisition protocol and not necessarily the best source for historical waveform data. Most SeedLink servers I have seen have less than 24 hours of buffer. Since you do have a fairly large buffer, I tried to see what I could do, but I am not going to be able to allocate much more time to this issue. There are ways to preserve the wave data for long-term storage that is more friendly for accessing historical data. Winston wave server works particularly well with Swarm. One option is to use Earthworm's slink2ew to import data into Swarm, and then to export it to Winston. If you are not familiar with these software, I can point you to some documentation that may help. You have my email address so please let me know if you would like this information.

[himmetgencer]

Please Can you send documentations to my email? Also I will reply to your comment next week.

Regards...