Session Manager Error

[mukoya]

Pacsone is not receiving images and times out on Echo, Query.

Immediately after service restart, is seems to work for a short while then gets locked.

The error in the logs is several lines similar to: "Mon May 16 14:43:59 2016 ERROR - SessionManager: Trying to delete session <128> while it's locked for I/O"

On Live monitor, I see items that sometimes do not disappear right away on manual delete.

I am routing to a network folder and there is an occasional timeout copying images to the folder, though this has been the case even before, but without the current error.

Any idea what could be the issue?

Regards.

[pacsone]

1. What's the version of PacsOne Server are you running?

2. The entries in the Tools->Live Monitor page can NOT be manually deleted at all, since they are system internals maintained by PacsOne Server.

[mukoya]

Thanks for your Quick reply.

Version is 6.4.5 64-bit

[mukoya]

The server worked fine eventually though I never quite figured out what the issue was.

I recovered it by copying WAMP and Pacone folders from a replication slave machine that is of identical setup and was working well. Now about 24hrs with no errors.

Additionally, I tested the issue of copy files to network folder by shutting down the destination server and thus making network folder unreachable. This did not result in the error either.

About never deleting an item on the Live Monitor page, I think the "Delete" button on the page can be misleading.

BTW, looks like a forward rule that involves copying received files to a folder does not create an entry on the Job status page. So, if folder is unreachable, you end up "losing" the files. Not sure whether Pacsone does any retries, but it seems the images are not copied when the folder later becomes available and there is also no way to manually retry.

Thanks.

[pacsone]

Currently PacsOne Server will log an error when copying to the destination folder fails, as retrying the file copy may not scale well when there are hundreds of thousands (or even millions) of routing jobs, not to mention the difficulty for scheduling the retries (e.g., when and how often to attempt the retries).

[mukoya]

Thanks for the info. I guess one then needs to keep destination folder reachable as much as possible.

BTW the error similar to "Mon May 16 14:43:59 2016 ERROR - SessionManager: Trying to delete session <128> while it's locked for I/O" is still occurring though.

However, it is just a few lines in the log file and appears not to be significantly affecting server operation.

Any pointers on possible source and solution for this error?

One feature pacsone can consider in future is to send mail on detecting certain text in the log, e.g certain error code. Can help one check behavior of server when error is encountered among other uses.

[pacsone]

These errors usually indicate PacsOne Server has timed out waiting for further packets from the peer AE/device, as it tried to delete the session object after the session has timed out. If you see a lot of such errors, then it means either the network link between the PacsOne Server host and the peer AE/devices is too slow or the peer AE/devices are not capable of handling multiple simultaneous Dicom associations with PacsOne Server. If it's the prior case, you should increase the network bandwidth of the network link, e.g., to 100/Gigabit/10-Gigabit Ethernet, etc. If it's the latter case, you should check the Dicom Conformance Statement of the peer AE/device and find out what's the Maximum Number of Simultaneous Dicom associations it can handle, then modify the "Maximum Outbound Connection" setting accordingly for that peer AE/device in the "Dicom AE" page of PacsOne Server.

[mukoya]

Thanks.
That is useful information.
It is probably bandwidth because all AEs connect via internet which is sometimes very slow.

However, it is probably more likely that something is freezing the server causing the timeout and the error is a result and not the cause of reach-ability problem. This is because during the time, server is not reachable even from localhost.

[pacsone]

1. Are you running the Windows or Linux version of PacsOne Server?

2. Can you consistently reproduce the problem?