Swsync

From CoolSolutionsWiki

-my other wikis

Contents

[edit]

Nov 4, 2008

  1. talk about open SRs
    1. Novell give updates
      1. get config from all servers, so I can run a report on bx2.lan
    2. SW give updates
  2. discuss options of NSL edir to AD
    1. no migration tools from edir to AD for NSL
  3. IDM
    1. discuss SW requirements for which attributes to sync and which data source is authoritative
    2. Unv Password for IDM and AFP
  4. new issues?
    1. Currently awaiting resources on ZCM DB in diff. time zones and ZCM DB best practices
    2. GW sent calendar items
  5. send out SR list prior call
  6. action items
    1. SW to sch onsite with symantec
    2. T to schedule onsite for IDM attribute and data autho. source
    3. procedure for migrating IDM 3.5 from lab to production. see wiki project
    4. research GW 8
      1. details on successfully running gw on windows
        1. GW IS supported on Windows, however GW is recommended to run on Linux
          1. performance is better, and stability.
      2. Win OS versions supported?
      3. verify GW 7/8 dependence on Edir
        1. YES GW 7/8 IS dependent on edir. I hear there are plans to change that in the future.
      4. client list of GW on windows customers
        1. PSE
          1. Karl K., Ron L Smith
    5. recommendations for centralizing PO.
      1. caching needed? t1 typical – some have frame relay of sharing 1.5 mb links
[edit]

Dec 2

  1. X Send SR report PRIOR to call
  2. review open SRs
  3. follow up with SW if they decided to implement nmas as a pilot or to use the pw self-service gadget (IDM project is awaiting SW's decision)
    1. yes on pilot nmas client
  4. TODO
    1. nmas health check for backend
    2. a lot of recent cluster abends
    3. TOP priorities is abends - awaiting logs from SW
      1. review last coredump - smoking gun???
    4. QSR observation on running GW on Win
    5. X send mandatory patches for nw, gw, edir, zen, client, nmas
    6. X find out if nldap abend fix is in 882 or 883 and send code to David
[edit]

Dec abend issues

[edit]

appears to be a loader abend - suspect msm.nlm, nlsrup, nlsmeter, connaud

  1. dec 8 abend is different from the one prior to updating msm.nlm, nlsmeter and connaud
  2. verify lvo-02 has auto tune ON -- the server.exe -u is ok
  3. IF 03 gets memory errors it could be set to the same as 02 for the server.exe -u
  4. TSAFS cache from 10% to 1% on both servers. IF they have memory problems
  5. option: disable multi-cpu
[edit]

abends due to migrating GW resources to another node and causing protected memory to abend server

  1. another customer began the troubleshooting process but has dropped the ball, we would like SW to troubleshoot this (including taking cores, applying updates as requested to resolve this issue)
  2. EIP in USERLIB.NLM
  3. may want to increase delay from 5 seconds 10 seconds - for the unload NCS script
  4. ACTION PLAN
    1. get coredump
    2. be willing to apply code fixes - last customer unwilling/unable to provide needed diag info
[edit]

galaxy abend

  1. customer has SR open with commvault
  2. does galaxy support sp8?
[edit]

loader SERVINST.NLM/snmp abend SR 10452279657

  1. Server LVO-02 halted Monday, December 8, 2008 2:17:24.466 pm
    1. A3C382D9 (SERVINST.NLM|(Code Start)+72D9)
  2. https://bugzilla.novell.com/show_bug.cgi?id=411021
  3. current: DS.NLM 8.8 SP2 SMP Version 20216.62 May 28, 2008
  4. fixed in: edir882ftf_2.zip
[edit]

SR 10453210785 Attempt to release an unacquired spinlock

  1. Dec 22 - send coredump to iscsi development team
  2. send fix Dec 23 - new iscsi modules (most from sp8, one fixed by dev) -- awaiting update from SW
[edit]

At least one of the nodes is Alive in the old master's node partition.

  1. LVC-GW4 halted Friday, January 2, 2009 2:32:33.122 am
    1. <paste> At least one of the nodes is Alive in the old master's node partition.

This node is NOT in the old master's node partition.

This is a split brain condition. Both sides of the split brain contain an equal number of nodes. The side that contains the previous master node wins. Nodes in the other side lose and eat this poison pill.

The Alive partition with highest node numbers should survive. This node is NOT in the Alive partition with highest node number.

This is a split brain condition. Both sides of the split brain contain an equal number of nodes but the previous master node failed or left the cluster. The side that contains the highest numbered node wins. Nodes in the other side lose and eat this poison pill. </paste>

  1. Recommendation: if this type of abend continues on a two node cluster-- a 3rd/2nd opinion to determine who to cast out is a good way to go- 2 nodes have a few more incorrect cast outs than 3 node and above.
[edit]

NLM unloaded without deleting all its processes.

  1. Server LVO-03 halted Thursday, January 1, 2009 6:19:25.464 pm
  2. need more info - what was the server doing at the time? anything unloading like galaxy? etc etc
[edit]

Jan 6, 2009

  1. Anyone going to gwavacon end of Jan?
    1. Gloria and TS are going
  2. review open SRs
  3. willing and able to troubleshoot and apply code fixes for memory address space issue?
  4. setup a time to do nmas backend health check
    1. hold off for a while
  5. how is the nmas client pilot going?
    1. hold off for a while
  6. TODO
    1. X info David which server this is/and get a config.txt--loader servinst abends with snmp on the stack - Dec 8 abend
    2. have TS show me problem user with GW multi-tabs
    3. X send follow up email and restate action plan to symantec
    4. X have Karen show NSL inheritance issue with newest code
    5. -> SR 10447384203 with Jacobson nsl still slow even after update 12 with iman 2.7 -- have Karen show issue and verify on correct code
    6. X send docs/training on migrating ZEN 7 to 10 and having ZEN 10 on AD
    7. find other customers that are using ZEN 10, particularly migrated zen 7 to 10 apps
    8. X send FST gina pw update FYI/INFO - perhaps get bug#
    9. X have David show his GW rule issue
    10. X resend recommendations on generic loader abend
      1. followed up with resources - stating we have an abend log with recommendations being followed, but no new coredump-please re-review coredump
    11. X Daron's ZCM 10 issues - emailed
[edit]

Feb 3, 2009

  1. GWAVA Con recap
    1. Gloria spoke with many folks - including presentor - said PO on windows corrupted DBs (small POs)
    2. significant feature upgrades
    3. GW 7.03 very stable
    4. Videos of Novell Executives using new GW 8 features for productivity
    5. GW has had a 3% increase with 30 million users
    6. PROJECTED
      1. GW 8 sp1 Aug 09
      2. GW 7 sp4 end of 2009
    7. GW Mon can restart GW agents if detected a failure (for sure on Linux...can't recall if on other plats)
    8. Teaming starter Pack (free 20 user teaming solution, those 20 seats still valid after purchase)
    9. GW Linux Best practices
      1. windows more likely to corrupt DBs, linux much less likely to corrupt
      2. Sles comes free with GW - as many sles servers as you need for your GW environment
      3. Ensure you bind exclusive to an ip addr, or Linux postfix will conflict with gwia/port 25



[edit]

Carried over from last month

  1. have TS show me problem user with GW multi-tabs
    1. scheduled for 1:30
  2. find other customers that are using ZEN 10, particularly migrated zen 7 to 10 apps
    1. Bryan T. can do this Tuesday or Thursday
[edit]

recap of GW product management

    1. SW needs an external calender - provided by GW 8
    2. request for enhancement (RFE): GW addr book for MS project
    3. Alex (GW PM): SWGAS is looking for a customer with similarly sized POs that are running on Windows 2008
    4. Alex: SW is looking for a customer with similarly sized POs that are running on Windows 2008
    5. Alex: SW' RFE "rich text in webaccess for tasks
    6. Alex: SW's RFE - smart delivery of large attachments for branch divisions - kind of like a Tiered elect distribution (TED)
    7. SW: quantify centralizing POs - they need to purchase 3rd party reporting tool in order to make a quantified decision.
    8. Alex/PSE: *** advise customer on if putting GW PO on a branch server that also has MS file/print is ok...or virtualize on vmware with sles GW 8--I say do a bake off and decide
      1. Option A: gw on win backend with all other MS services
      2. Option B: virtualize GW on vmware ...running sles and gw 8 --I'm not sure I like this...if it was xen I KNOW that works!..but I don't have experience with GW on vmware...other than it used to NOT be supported and DBs don't like to be virtualized
      3. Option C: pull POs to the division level...but that introduces complexity like 20 users requesting the same file 20 times over the WAN...but if they do caching and set the max size to cache at 1mb...that may help...but then they'll complain about not being able to see delivered messages while in caching mode
      4. Option D: put the PO on MS 2008 at the branch level as they are small and should be "ok", but keep the large POs on linux -- I think this may be my favorite solution -- Alex...thoughts???
    9. Alex: SW is expecting recommendations on running GW on Windows etc etc...
[edit]

Feb 5, 2009--discuss Bryan's ZDM 7 apps to ZCM 10 migration experience

  1. mostly user distros
  2. RC
  3. app distro -- snapshot/msi? older apps are snapshot, newer are msi/install shield. Most msi have command line switches
  4. 400-500 apps need to be migrated
  5. big motivator for zdm 7 to zcm
  6. using zen for launcher/shortcuts
  7. DLU
  8. patch link...outside zenworks
  9. WAN links - t1
  10. 4000 computers (gobooks, desktops, laptops etc), 2500 users
  11. use significant variables for Novell Login scripts -- apps use a ton of application variables.
  12. run once and quid inator (cleans up quids to redeploy)
  13. bumping the number in ZCM 10 is not as "impactful" as zdm 7
  14. biggest gotcha is figuring out how to deal with app migration
  15. 8 year old apps, with macros and snapshot
  16. suggest use case -- lots of testing!!!
  17. RC policy that requires user accept remote control--we can uncheck require permission from user.
[edit]

Mar 3, 2009

  1. attendees
    1. Mike O, TW, Gloria, David, Louis
  2. review open SRs
  3. review TODOs
  4. review action plans
  5. thoughts on a win2008 server running vmware with a sles server running GW
    1. experts don't always agree, I would get a statement from a GW Product Manager.
  6. encrypt email
    1. secure and easy-- protocol or message level, secure.
  7. abend are still main issue
[edit]

May 5 2009

  1. attendance:
    1. Louis, David, TS, Gloria, Mike?
  2. update on IDM project
    1. pw sync
    2. Group
  3. Open SRs
    1. discuss NSL Oracle
  4. New issues?
[edit]

TODOs

  1. Dup multi-tab user issue
  2. Continue working IDM project
  3. X contact Bill J. and get dev teams together.
    1. left vmail and emailed Bill J.
Retrieved from "http://wiki.novell.com/index.php/Swsync"