Weird conquest server hang on virtual query

  • Hello,
    I've started a new topic since we have encountered a new weird error with our conquest PACS setup.
    The issue only occurs when trying to retrieve images from two virtual servers which hold our images from a Q/R server that is accessed from all the workstation in our institution.
    Basicaly, what happens is that when we issue a retreive command from the q/r server from the two virtual servers, sometimes, absolutely arbitrarily during the transfer the q/r server hangs, with no evident error message (we even used debug level 4). We had to use keepalive command in order to make the q/r server working.
    The odd things are:
    1) It only happens to about half of the studies we try to retrieve and only to those.
    2) Judging on 1) we concluded that those studies might be corrupted (although they were previously imported via a c-move command into a conquest server residing on the virtualserver, and thus considered clean). However, the even more bizarre thing is that if we move the images via a manual retreive command issued on the q/r server from the specific virtualserver (no longer considered a virtualserver therefore) the study is received with no error. Studies being corrupted or incompatible of somehow virtually is ruled out as a cause.
    3) If we do 2) and bring manually the studies in the q/r server, if we remotely query/retreive that study, it sends it with no problem (because of cache, we reckon)
    4) The server has no compression set on it (plain ol' dicom) and receives the files as *.dcm
    5) The virtualservers store NKI .v2 files, and decompress them before sending. However, they also issue absolutely no error upon sending the files, they just ... stop sending... as if the q/r server were down (which is the case, actually)
    6) We have diagnosed the network and ruled out as a cause. So has been the hardware on the machines. The error happens at a random image in the study that is being sent. Not necesarily the same. But always on a particular study that is known to fail this way.
    7) Windows event log says the dicom server service stoped for "acces denied" which seems to be a rather unspecific error.
    8) We've tried different types of configuration on the q/r server regarding virtualcache, perimage value etc.
    Any suggestions?
    Thanx,
    Radu

  • Hi,


    No clue yet. did something change when this started? What version(s) and databses are you using? And can you sketch out your setup and where is hangs?


    q/r server --> virtualserver --> pacs


    Marcel

  • Hello there,
    We are running Conqeust 1.4.15 on Win XP x86 and Vista x64 (on one of the storage servers). We use MySQL (native) on both storage server, and played with dBase and MySQL on q/r server (trying to rule out database issues on that comp).
    The chain is as follows:


    Workstation(s) <-> q/r server <-> Storage (2 virtualservers)


    Briefly, when we need an exam, we make a query from workstation to q/r server which in turn virtualqueries the storage servers. They respond, send data to q/r server which sends them along to calling workstation.
    The hang is when the storage server moves data towards the q/r server. In response, the workstation gets nothing, of course.
    The error is usually "Association lost" and a dead q/r server (which must always be restarted manually or keptalive in order to continute working). The storage servers are fine, ending their c-move threads with no errors.
    One more weird thing that might be important for you is that the database gets corrupted. Meaning it has the entries for the files but not the actual files on the mag device. So, when we try to remove the files from the q/r database manually, it gives an error as it is trying to remove non-existing files. It doesn't hang or anything, and it deletes the entries from the database.
    Sometimes, not clear when, it gets the files, puts them into database and disk, and hangs when it would be supposed to send them over to the quering workstation. When that happens and I try to delete the patient from database, the output error is repeatedly:

    Code
    [CONQUESTQR01] Removed file: [MAG0.1:1570306343226\1.3.46.670589.10.1.1.2163771294.1217317526.853730_0001_000001_12583321200183.dcm]
    [CONQUESTQR01] ***Could not remove IOD C:\conquestqr01\Data\1570306343226\1.3.46.670589.10.1.1.2163771294.1217317526.853730_0001_000001_12583321200183.dcm


    It's indeed quite tricky and ambigous.
    Digging deeper, seems like the error is somewhat computer related. I've emulated the q/r server on my laptop and it worked like a charm.
    I changed the server configuration as much as I could trying to figure out what is causing this (switching from MySQL to dBase, even reinstalling conquest from scratch), couldn't find a clue. Played with virtualcache 0/1 etc.
    It might be ethernet related, windows related (access denied is a weird error for this service).... whatever. Will probably reinstall windows and try again.
    In you forever debt,
    Radu

  • Hi,


    during a virtual query, the image list is first stored in the database with device ID, e.g, MAG0.1. Afterwards the data is collected from VirtualServer1 to MAG0 and then transmitted. So if the collect from VirtualServer1 fails, device ID's like MAG0.1 stay in the database. So that part of the operation seems to go well. It then scans the list and builds one or more c-move requests. Did you try displaying a debug log on the Q/R server? ANy other differences between the laptop system and the q/r server (both 32 bits?).


    You may also try to downgrade the q/r server (temporarily) to 1.4.14 (just replace the dgate.exe) to see if the problem was a recently introduced bug.


    Good luck,


    Marcel

Participate now!

Don’t have an account yet? Register yourself now and be a part of our community!