Bug 523384

Summary:	During /var/satellite conversion, you can run out of UNDO_TBS
Product:	Red Hat Satellite 5	Reporter:	Miroslav Suchý <msuchy>
Component:	Upgrades	Assignee:	Jan Pazdziora <jpazdziora>
Status:	CLOSED ERRATA	QA Contact:	Brandon Perkins <bperkins>
Severity:	high	Docs Contact:
Priority:	high
Version:	530	CC:	cperry, dherrman, jhutar, jpazdziora, mzazrivec, pep, psklenar, tao, tscherf, xdmoon
Target Milestone:	---
Target Release:	---
Hardware:	All
OS:	Linux
Whiteboard:
Fixed In Version:		Doc Type:	Bug Fix
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2009-10-06 14:36:01 UTC	Type:	---
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:
Embargoed:
Bug Depends On:
Bug Blocks:	518256, 523386

Description Miroslav Suchý 2009-09-15 09:51:13 UTC

Description of problem:
in rhn-upgrade documentation in step 3 we say to run:
 /usr/bin/update-packages --db=dbusername/dppassword@dbSID
This is run in one big transaction, which will require a lot of UNDO_TBS.
Additionaly if we run out of the space, the whole transaction is rollbacked, but some files is already moved and stay moved, which will introduce inconsitency between paths stored in db and path on file system.

We should commit after each package/path move. But this will by probably slow, so due speed reason we may want to commit after let say 1000 moved packages.

Version-Release number of selected component (if applicable):
sat530

How reproducible:
during upgrade

Steps to Reproduce:
1. upgrade sat520 to sat530
2. during step 3 watch UNDO_TBS grow.
3. optionaly make sure you run out of UNDO_TBS in middle of transaction
  
Actual results:
slow and mess in db

Expected results:
fast and correct behavior.

Comment 1 Jan Pazdziora 2009-09-16 14:13:32 UTC

Taking for now.

Comment 2 Jan Pazdziora 2009-09-16 14:25:02 UTC

Fix to break up the transaction into smaller chunks so that we don't run out of the rollback space committed to Spacewalk master: d00dc3a5efd28f683fd680cabffb7ae969895900.

Comment 3 Jan Pazdziora 2009-09-16 14:27:22 UTC

Note that the processPackageKeyAssociations function does commit, so if most of the packages have keys (?), during the first run (when the key records are being added to the database), the loop behaves as autocommit anyway.

Comment 4 Jan Pazdziora 2009-09-16 14:45:52 UTC

Not undo_tbs related, but potentially useful changes: d4d824d468b2b682f7bc02b9f432fa5c75f5e5d6 and 714f2d07cbbd8e19ddde53f4e5a80c33b5f137ff, to provide small speedups.

Comment 5 Milan Zázrivec 2009-09-22 12:34:11 UTC

satellite.git, SATELLITE-5.3: 18585b473f0e69af1704d2557c1aa946ff4a30a1

Comment 8 Milan Zázrivec 2009-09-25 07:41:15 UTC

rhn-upgrade-5.3.0.24-1.el4sat & rhn-upgrade-5.3.0.23-1.el5sat

Comment 14 errata-xmlrpc 2009-10-06 14:36:01 UTC

An advisory has been issued which should help the problem
described in this bug report. This report is therefore being
closed with a resolution of ERRATA. For more information
on therefore solution and/or where to find the updated files,
please follow the link below. You may reopen this bug report
if the solution does not work for you.

http://rhn.redhat.com/errata/RHBA-2009-1479.html