1 2 Previous Next 24 Replies Latest reply on Jun 27, 2011 8:54 AM by mrspike

    Sdstatus folder on core server


      On our landesk 9.0 core server folder "C:\Program Files (x86)\LANDesk\ManagementSuite\Ststatus" contains huge amounts "*.swd.xml" files. I cannot browse this folder. Do you someone know which service make these files and how i can stop this.


      Thank you for advice.

        • 1. Re: Sdstatus folder on core server
          phoffmann SupportEmployee

          *Moving thread to software distribution, since it's got nothing to do with installs*


          Those files are client-side updates (Software Distribution Status) - and they should be getting processed.


          First of all - how much exactly are we talking about when you say "huge amounts".


          Secondly - these should be getting processed ... is your Core just busy with other stuff? Are the files getting processed at all (i.e. - do older files get processed, and you just keep getting lots of new files in?). How many nodes worth of environment are we talking about here?


          How heavily stressed is the server just based on CPU and memory?


          This is not the sort of thing that shoud be attempted to be resolved by stopping services.


          - Paul Hoffmann

          LANDesk EMEA Technical Lead

          • 2. Re: Sdstatus folder on core server

            OK, I moved thread to "Software distribution".

            "Huge amounts" means more than 1 500 000 files.But now I can not browse this folder on core because if i click on it explorer "not responding", therefore I cannot load content , and can not see if some file are processed.


            CPU an RAM on the core is in normal I think.


            I see in task manager on the core there is open a lot off postcgi.exe*32 processes (maybe 300) , before I have never seen this.

            • 3. Re: Sdstatus folder on core server
              phoffmann SupportEmployee

              What I meant with "moving..." is to say that *I* had moved the thread.


              Second, these bits of data are still outstanding:

              - How many nodes (roughly) are we dealing with here in your environment?

              - What are the specs of the Core Server (# of CPU's, how much memory)?


              'Thinking' that things are normal is not sufficient here - We need to be sure. So - is the CPU maxed out or is it not? How much memory is being used here (and how much is still free)? This needs to be checked.




              Third - 1.5 million files ...? Well, that's not surprising that the OS is going into problems with that, Windows usually runs into problems with anything over 65,000 ... rather than using explorer.exe, use command-line stuff. Old DOS-type commands are much more helpful in this sort of situation.


              Let's just make sure how much we're talking here... so do this.

              3.a - START -> RUN -> "CMD"

              3.b - go to the SDSTATUS directory.

              3.c Run:


              dir /w


              and see what the counts come back with - I'm hesitant to believe that there's really 1.5 million files without making sure... one and a half million status updates don't just get generated over night...


              The fact that there's 300-odd POSTCGI processes is a bit worrying, what exactly are you running as a distribution job? If there's nothing you're running, sounds like something's been misconfigured on the clients for them to be pounding the Core this hard.


              More context / environmental information would help a lot.




              - Paul Hoffmann

              LANDesk EMEA Technical Lead

              • 4. Re: Sdstatus folder on core server

                We have roughly 6000 devices, and new version LDMS 9.0 is running 2 moths , we migrated it to new server from ver.8.8 SP3.
                Server has got 4 x Intel Xeon 3 GHz procesors and 6 GB RAM. CPU is not maxed, and now it is average busy at 50 %.
                This problem maybe runs longer time, but I noticed it only yestarday.

                Dir command from sdstatus directory listed me 1 485 515 files, screen is in attached file (executing dir coomand took 20 minits) .

                POSTCGi procceses now not are running, this situation from my previous post was after restart server.


                About our distribution job ? We use some regluar tasks with delivery method : police ,police-push method,push.
                But few task we have configured with delivery method "always listed for intsallation" that are asociated to our all devices and they are sheduled to run aevery day.


                I attach one of these files fro sdstatus directory.

                • 5. Re: Sdstatus folder on core server
                  phoffmann SupportEmployee

                  Wow ... that's very much a first.


                  So let's see ... 1,5 million % 6,000 (nodes) == 250 XML's per client.

                  250 / 60 (days) == 4


                  ... so that's circa 4 status updates per client per day.


                  In and of itself that doesn't seem to be terribly bad... hummm ...


                  All right - well, let's give your Core a leg up, first of all. First order of the day and all that.


                  1 - Stop all Intel/LANDesk Services on the Core. Most importantly here are LANDesk Inventory, Scheduler and Policy services, but 'to be on the safe side' let's stop all of them.


                  2 - Rename "SDSTATUS" to "SDSTATUS.OLD" and create a new directory called SDSTATUS (make sure that the NT- permissions of the new directory are the same ones as the ones on the "old" one. You can then move the directory on to another box, another hard-drive if you want to retain it, or delete it if you're OK with potentially loosing that data (policy updates).


                  3 - At this point, make sure you've got the rollup-patch installed for 9.0 ... whatever's causing your problem is likely to be related to policies, but may as well make sure the bases are covered.


                  4 - Reboot the Core. In case there's something "Microsoft'y" that's hung or whatnot, let's give the gal a full reboot, see if it'll behave better.


                  5 - after the reboot, keep an eye on the 'new' SDSTATUS directory, getting the XML-files in there is a normal part of policy processing, but they should be consumed. Potentially running something like FILEMON on the directory may help you (you want to filter for errors), in case there's anything like NTFS permissions or whatnot that's causing you problems in having the files be deleted.


                  ... let's begin with that, and go from there.


                  ... also - have a look at the APMSERVICE.INI (that's in the "C:\Program Files\LANDesk\ManagementSuite\" directory on the Core) - is there a mention of:





                  in there? That would explain what's been going on (this can be enabled for debugging - but for reasons that are rather obvious to you at present, should not be left on indefinately) .


                  Let's see how this will get you moving along. But very much an unusual thing to say the least .


                  - Paul Hoffmann

                  LANDesk EMEA Technical Lead

                  • 6. Re: Sdstatus folder on core server

                    I did it and now it is OK

                    files are added up there but they are "consumed", it works now correctly.
                    From until now i will keep my eyes on this folder too.
                    Thank you for your help Paul.

                    • 7. Re: Sdstatus folder on core server
                      phoffmann SupportEmployee

                      Well - that's good to hear .


                      Not sure what must've gone wrong for it to happen to begin with, but once Windows has several 100,000 files in a directory things are more likely to get worse rather than better, with the OS hanging itself trying to just access them.


                      Glad to hear that this is resolved for now then . That should be one more well behaved Core out there now .


                      - Paul Hoffmann

                      LANDesk EMEA Technical Lead

                      • 8. Re: Sdstatus folder on core server

                        its interesting how I have the same exact issue. Server is being given all the space (its a VM so I am growing Disk) as soon as it is given. If i run any tool, it just hangs on the folder sdstatus.

                        Completing a reboot after changing the name of the folder, will know in a minute if this resolves some of the issues i am having with the core. Definitely is the same issue as yours that i am having. Opened a ticket but never heard anything back about it... just a general shrud along with.. "clean your temp files".


                        After the reboot, it didnt create a new folder by the name of sdstatus. Stopped the services and deleted the folder altogether. Restarted server again...


                        well.. i am going to delete the contents of the folder through command line in PXE environment and see how it goes. I'll post back and update status.   

                        • 9. Re: Sdstatus folder on core server
                          phoffmann SupportEmployee

                          Urm - given that it's a www-service that's trying to throw the files into SDSTATUS, renaming/deleting the folder is a very bad idea.


                          I would rather recommend that you re-create it ASAP, and give it the permissions it should need (check your test Core), otherwise you're not going to be getting any updates at all.


                          At a push, moving the contents into another directory is fine (since it's just files, you can copy them back in batches of a few hundred as needed, ot be processed) ... but generally, never delete a directory unless you're absolutely positive it won't do any harm.


                          - Paul Hoffmann

                          LANDesk EMEA Technical Lead.

                          • 10. Re: Sdstatus folder on core server

                            Got in the office at around 10 am and it was still deleting the contents of the folder. I cancelled the task and rebooted the server. After getting back in the server, files (over 1000000) started to go down slowly but then more files were getting dumped in it. moved all the files off that folder and cleaned it up. Seems like the server is running much better now that it has white space as well as it has caught up to speed.


                            Paul, you mentioned in your point 2 to rename and create another one but you replied to my post and said not do that.

                            • 11. Re: Sdstatus folder on core server
                              phoffmann SupportEmployee

                              Point 2 where?


                              Ah right - there, 3 replies above, I see.


                              Yes, renaming works as long as you re-create the relevant directory. The thing I was 'protesting' about was "just" renaming the SDSTATUS folder, without putting a replacement for SDSTATUS in (as we then couldn't get any software distribution status updates).


                              ... did you really have over 1,000,000 (1 million) files? If that's so, no wonder things went slow. Windows tends to get REALLY bad with dealing with more than 65,000 files. This becomes then a self-aggravating problem ... the more files there are, the slower Windows gets in getting a directory list / accessing any file (so the service processing them takes longer and longer on each single file), while still being able to write files into that directory.


                              For reasons I don't quite understand at this point, it's - apparently - able to still write files to the directory when this happens, just that pulling the directory contents becomes a pain in the back, never mind grabbing files.


                              You have two choices here really, in such situations. Delete all the files (via command-line, no way you want to do this through the UI), or move the files and re-inject them into the SDSTATUS folder in batches (of - say - 1,000 or so).


                              I hope this clarifies things a bit? Apologies for any confusion I inadvertently seem to have caused.


                              - Paul Hoffmann

                              LANDesk EMEA Technical Lead

                              • 12. Re: Sdstatus folder on core server

                                Well, server has been running quite well after removing all the files from teh sdstatus directory. I did try to delete the folder after booting up the server in PXE so that it does not keep making more files. When i tried removing the files while in window, it was generating more files than it could delete... honestly.


                                So, in PXE, tried deleting the folder and left it over night. Read your post and you suggested not to delete the folder. restarted the server and tried deleting all the files in it. Once all the files were gone, server was running Much better and i got back about 40GB of space. It is kind of bad when the server takes up this much space with 1KB files. 


                                Everything is working after the workaround. I dont think you can delete all the files in there while Windows is running. Also, you Have to be Patient while deleting the files through CMD as it Does take forever (at least 7-8 hours).

                                • 13. Re: Sdstatus folder on core server

                                  Again after update Landesk 9.0 with SP2 sdstatus does not  catch consum incoming xml files. Tempo of increasing is faster as decreasing. During approximately 1 hour there is 1500 files.


                                  Content these files look like this :
                                  <?xml version="1.0" encoding="UTF-8" standalone="yes"?><alert><taskId>1372</taskId><packageId>0</packageId><complete>false</complete><retcode>229392385</retcode><deviceid>{D4E2BB37-2214-AC4A-8B66-1A560FF3B531}</deviceid><ldapinfo>CN=X Y,OU=X,DC=Y,DC=YY,DC=ZZZ</ldapinfo><message/><logfile/><alert_ostype>windows</alert_ostype><alert_osfamily>winnt</alert_osfamily><alert_pds2id>37BBE2D4-1422-4AAC-8B66-1A560FF3B531</alert_pds2id><alert_inventoryid>{D4E2BB37-2214-AC4A-8B66-1A560FF3B531}</alert_inventoryid><alert_host>PCName.X.Y.ZZZ</alert_host><alert_state>1</alert_state><alert_localTime>2010-12-10 12:55:36.749</alert_localTime><alert_gmtTime>2010-12-10 11:55:36.749</alert_gmtTime><alert_instance/><alert_id>internal.100_10_10_24.swd_sdclient.status</alert_id><alert_ruleset name="SoftwareDistribution Ruleset" date=""/><alert_health>2</alert_health></alert>

                                  • 14. Re: Sdstatus folder on core server

                                    Kapak - you need to contact support and request a couple of patches possibly as we have been fighting this all week and other customer too from what i've heard.  you need a patch for the APM service not keeping up with all the task status xml files being sent to the sdstatus directory and you may need a patch for the clients as well for the "task queue storm" where they can get into a state of sending way too many status alerts to the core sdstatus directory.  when we came in on monday this week, we had 1.3 million files in sdstatus before applying any hotfixes/patches.  we are on ldms 9sp2.

                                    Thanks and hope this helps you.

                                    1 2 Previous Next