Server frisst Speicher

Die Knowledge Base sagt folgendes dazu:

Zitat

[size=xx-large]Compact Task Allocates All Memory and Crashes the Domino Release 5.0.1 Server[/size]

[size=large]Abstract[/size]
After upgrading a Domino server from Release 4.x to 5.0.1, the first time the Compact task is run, it allocates all available memory on the system until the server crashes. The issue does not occur when compacting all databases, just certain ones. You have not been able to determine what causes the server to crash on these certain databases. You can monitor the problem if you do a dir command for the database name, without specifying an extension. You see the NSF file and the TMP file. The TMP file grows larger and larger. dir db1*
db1.nsf 1,000,000 Bytes
db1.tmp 5,000,000,000 Bytes You can also start the Windows NT Task Manager and watch the memory allocation of the NCompact task.

[size=large]Content[/size]
This issue was reported to Lotus Quality Engineering and has been addressed in Domino Release 5.0.2.

Release 5.0.1 has trouble converting the ODS to the new version 41 (this happens on the first time Compact is run on R5). If you try to create a new replica of these databases from a 4.x server on a 5.x server, the system also hangs (when creating a new replica on a 5.x server, the ODS also converts to 41).

Workaround #1:

Find the problematic databases with trial and error; testing your production databases on a 5.0.1 test environment. You can use a batch file (send a dir command to a txt file) to save time.

ncompact d:\notes\data\db1.nsf
ncmpact d:\notes\data\db2.nsf

After each crash, customize your batch file to start with the next database and not from scratch. Create a new replica with an .NS4 extension on your production system (Note: simply renaming to an .NS4 at the operating system will not prevent Compact from upgrading the ODS). The 5.0.1 Compact task will compact the databases but will not try to convert the ODS to 41. Everything works fine.

Workaround #2:

Migrate your 4.x system to 5.0 or 5.0a and run the Compact task on all databases. Conversion to ODS 41 works fine for all databases. The problem only occurs with 5.0.1. Upgrade from 5.0 or 5.0a to 5.0.1.

Supporting Information:

The crash/panic could occur on any server task because of a lack of memory. It is not always the Compact task. The issue also occurs with 5.0.1a.

Workarounds which do NOT solve the problem:

1. Fixup, Updall, Compact, New Copy or New Replica for these databases on a 4.x or 5.x server.
2. Copy NCompact.EXE from 5.0a to 5.0.1. The problem is in the kernel.
3. Create a new replica from a 4.x server with a 4.x or 5.x client on the 5.x server.

Alles anzeigen

Wie oft wird denn der Compact-Task angestoßen? Täglich?
Ggf. wird der Compact in einer Nacht+Tag nicht durch und dann wird in der nächsten Nacht der nächste Compact gestartet.

Evtl. die 17-GB-große DB aus dem nächtlichen Compact rausnehmen und nur am Wochenende komprimieren.

Zitat

DrNotes wrote:
Würd mich interessieren auch welcher Knowledgebase?
Kannst mir vielleicht den Link schicken?

Das ist ne DB, die von IBM runterrepliziert wird. Nicht öffentlich. Deine Firma muss sich registrieren und ich glaub auch Schotter abdrücken

Die KnowledgeBase wird (zumindest bisher) mit den Passport-CD's ausgeliefert und zur Verfügung gestellt. Lediglich der Update-Service per Replikation mit IBM ist kostenpflichtig.

Die KnowledgeBase hat einen öffentlichen Teil, einen Businesspartnerteil und einen IBM-Internen Teil.

Der öffentliche Teil ist für jedermann per Browser hier zugänglich:
Support - IBM Public KnowledgeBase