In a previous post, I talked about the Conquest server slowdowns we have been experiencing on an increasing frequency. Previously, I thought these problems might be related to WAN network issues.

Last week I made another discovery, this one I suspect being the culprit. Twice last week I was able to respond to calls about slowdowns immediately, and rather then just restart Conquest to fix the problem, I spent some time looking into it. Auto-forwarding of images over the WAN had been turned off, but since the problem continued, it did not seem to be the cause of the problem. We also auto-forward images from Conquest to several eFilm workstations to expedite review. Upon closer inspection of one of these eFilm PCs, I noticed that its hard drive was full, or at least, it had reached the threshold eFilm has for maximum drive space usage, so eFilm was in the process of deleting images. It would seem that eFilm is not very good at multitasking, as it was severely slowing down the receiving of files for this review station. On the eFilm computer, I killed the DICOM server service, which was behind in receiving images from Conquest. Immediately, Conquest sprang back to its regular pace, and started receiving/sending images in a quick manner.

So now that I have apparently narrowed down the issue to eFilm and disk space, I still need a way to automatically manage it. The eFilm process that manages disk space is called Disk Manager, and its exe is DM.EXE. This file has some command line parameters, but I was unable to get them to do much. I was hoping to run it after hours each evening, so it would have lots of space for the upcoming day, but had no luck.

What I did setup was a process using PSTOOLS, specifically PSEXEC, so that when Hostmonitor (our monitoring tool) detected space issues on a PC, it would run a command freeing up space by deleting the oldest studies from the eFilm workstation’s hard drive. So far this week, this seems to be working, and we have not yet had a slowdown. However, this solution is flawed, as we are not removing the images from the eFilm database, rather just the actual files on the hard drive. There are two consequences to this. Firstly, a user will still see the images listed in eFilm, and will be unable to open them. Secondly, and more important, the eFilm database will keep growing, and it has a fairly small threshold before it develops issues because of being too large.

Today I thought of perhaps a better method of dealing with this issue. Instead of deleting files off the PCs, I am going to add files, or specifically, a single very large file. How will adding files help with disk shortage issues you may ask?

I plan on letting eFilm’s Disk Manager do what it is supposed to do, and delete images. When I copy this large file over (after hours), eFilm will notice that the drive has passed its threshold of max disk usage, and will begin properly purging files until the predefined disk space becomes available. At which point, I will simply have this large file delete itself, and presto, lots of free space for the next business day.

This should be put into action by the weekend, and I will post the results.

Bookmark to: