Pegasi Wiki

This wiki acts as a memo for our own work so why not share them? Feel free to browse and use out notes and leave a note while at it.

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

edirectory_obituaries [2016/08/17 11:35]
Pekka Kuronen
edirectory_obituaries [2017/11/06 10:11]
Line 1: Line 1:
-===== Fix stuck obituaries in eDirectory - quick step by step guide ===== 
- 
-You may get error "​Error:​ -637 "​Previous move in progress"​ when IDM is not syncing properly or perhaps when trying to do a partition operation such as server removal. This is most likely a stuck obituary or obituaries which need resolving. Read on. 
- 
- 
-==== For all stuck obituaries ==== 
- 
-Firstly see that time is in sync, replicas in sync, do a filesystem check if needed, reboot etc. 
- 
-  * ndsrepair -T 
-  * ndsrepair -E 
- 
-Time must be in sync and errors must be 0. 
- 
-Check stuck obituaries: run this locally in all of the servers: 
- 
-  * ndsrepair -C -Ad -A 
- 
-If you see stuck obituaries (BACKLINK, DEAD) try to nudge backlink process by firstly: 
- 
-  * ndsrepair -R -Ad -XK3 
- 
-and secondly starting up ndstrace: 
- 
-  * ndstrace 
- 
-and in the ndstrace console issuing commands: 
- 
-  * set dstrace=*b 
-  * exit 
- 
-You can also try and re-timestamp the obituaries so that eDirectory will process them again 
- 
-  * ndsrepair -R -Ad -A -OT 
- 
-If you see older obituaries they may not get processed by the commands above so here's how to re-timestamp old obituaries 
- 
-  * start ndstrace 
-  * type set ndstrace = *j 
-  * type set ndstrace = *f 
- 
-Give it a little time and check again with "​ndsrepair -C -Ad -A" 
- 
-Also another mind trapping jedi-ndstrace-trick with obituaries is: 
- 
-  * set ndstrace = off 
-  * set ndstrace = nodebug 
-  * set ndstrace = on 
-  * set ndstrace = +sync 
-  * set ndstrace = +jntr 
-  * set ndstrace = *u 
-  * set ndstrace = *h 
-  * set ndstrace = *j 
-  * set ndstrace = *f 
- 
- 
-==== NON orphaned obituaries that just don't go away ==== 
- 
-This is the weirdest piece of information I've come across for a while. Some things just are not printed or I just cannot read them. 
- 
-If you have MOVE and INHIBIT_MOVE obituaries nicely in pairs that just stick around (unprocessed status etc) and will not go away no matter how many Novell / NetIQ manuals you read try this. 
- 
-    * Go to iMonitor 
-    * Fire up advanced mode (look above) 
-    * Go to reports and report obituaries (look above) 
-    * Click on an "​Advanced options"​ of the obituary 
-    * Add '&​op=all'​ to the end of the URL 
-    * Select "​advanced options"​ 
-    * Select "Purge xxx (purgeobj)"​ 
-    * Observe your nerves relieving as those obituaries fly fly away.. 
- 
- 
-==== For orphaned INHIBIT_MOVE obituaries ==== 
- 
- 
-For stuck obituaries which are mostly inhibit_moves WITHOUT corresponding MOVED obituary you can release them with iMonitor. I remember to have done it to DEAD obituaries as well but I am not sure. Better google first :) 
- 
-Here is how to proceed: 
-  * Locate the orphan obituaries with command (for each server if you have more than one in your tree): ndsrepair -C -Ad -A 
-      * For easy observation search or highlight test "​MoveObit:​ destID"​ and see it following legit INHIBIT_MOVE - MOVED pairs. What we are removing are the orphaned ones, not the paired ones. 
-  * Connect to iMonitor (for each server): http://​server:​8030/​nds ​ 
-  * iMonitor -> main page -> Advanced mode enabled 
-  * iMonitor -> Reports -> Report Config -> select obituary listing configuration 
-    * obituaries older than "​1"​ hour  ​ 
-    * Include obituary types: restored, dead, moved, inhibit move 
-    * Run report 
-    * Locate the obituary you identified with ndsrepair 
-    * Select "​release move entry" and be sure you have the right obituary! 
- 
-==== Other useful ndstrace commands ==== 
- 
-  * set dstrace=*h : ​ force resync 
- 
- 
-==== For really really bad problems with your flaim database ==== 
- 
-Open ndstrace and look, initiate sync between servers if needed and observe the hex addresses of the objects that are broken. Then use "​ndsrepair -J <​hex_address>"​ to fix the single object using the entry ID's hexadecimal pointer. ​ 
- 
-This can take quite a bit of time but I managed to salvage one really messed up eDirectory with a few hours of manual repair work :-) 
- 
- 
-Also check https://​www.novell.com/​support/​kb/​doc.php?​id=7004722 
- 
-{(rater>​id=edirectory_obituaries|type=rate|trace=ip|tracedetails=0|headline=off)} 
- 
-~~DISCUSSION|Leave a comment~~ 
  

  //check if we are running within the DokuWiki environment if (!defined("DOKU_INC")){ die(); } //place the needed HTML source codes BELOW this line