Outstanding issues for 3.0

gLite 3.0 released

26/05/2006: This issue tracking page is no longer updated, please see the EMT Home Page for details.

Problems found in RC4

STATE

• OPEN - not fixed.
• FIXED - understood and with an fix available.
• CLOSED - fix included in release.

Number	Component	Description	Savannah	Status	State	Action
105	Yaim	DPM_FILESYSTEMS in site-info.def example of YAIM	#16492		FIXED	In Yaim head
106	glite-CE	gliteCE configuration needs to be run twice after a fresh install	#16496		FIXED	glite-CE version 2.4.7
107	Yaim	Confusion in meta pkg name	#16509		FIXED	Yaim head
108	LCMAPS	bulk submission fails when users' DN mapped to single account	#16295		OPEN
110	FTS	BDII plugin sometimes crashes or abort when listing associated services	#16510		FIXED
111	FTS	BDII plugin returns no entries when quering by host and type	#16505		FIXED
112	FTS	shouldn't have to specify default logging params in YAIM	#16521		CLOSED	Not Critical
113	Dcache	gsidcap fails when > 56 CAs on server	#16538		OPEN

Packages required for RC4

• RC4 must contain any relevant LCG-2_7_0 updates
  • http://linuxsoft.cern.ch/LCG/apt/LCG-2_7_0/sl3/en/i386/RPMS.lcg_sl3.updates/
• New version of Yaim from HEAD
• pnfs-3.1.10-15 needs to be added to the dCache meta-package
• glite-config-1.6.29-0 package
• lcg-info-dynamic-scheduler-generic-1.5.2-1
• lcg-info-dynamic-scheduler-pbs-1.5.1-1

Problems found in RC3

Number	Component	Description	Savannah	Status	State	Action
97	Meta Package	glite/lcg CE targets confusing for configuration and installation	#16385 (16356, 16416 are dupes of this) Apr/25 fixed on yaim HEAD	Nasty Yaim Gotcha.	CLOSED	glite-yaim-3.0.0-9, checked by Di
100	FTS	BDII publication does not work properly	#16453		FIXED	Yaim head
101	DPM	YAIM starts rfio before creating...	#16465		CLOSED	glite-yaim-3.0.0-9, checked by Di
102	dCache	meta package glite-SE_dcache lacks...	#16463	pnfs-3.1.10-15 needs to be added to the dCache meta pakcage	CLOSED	RC4, checked by Di
103	Yaim	Rationalizing Yaim targets	#16466		FIXED	Yaim head
104	Tank	MySQL configuration	#15656		FIXED	Yaim head
109	VOMS	Upgrade fails for VOMS Oracle	#16475		FIXED	RC4 Maria to check

Packages required in RC3

• glite-security-config version 1.2.4
• glite-WMS version 2.3.3
• glite-iperf-1.3.2-1.i386.rpm
• glite-udpmon-1.3.10-i386.rpm
• glite-CE-2.4.5-2
• patch #700: condor-lcg-1.1.0-1
• glite-yaim-3.0.0-6.noarch.rpm
• patch #710: LFC / DPM 1.5.6

RC2 Problems

Found in Certification

Number	Component	Description	Savannah	Status	State	Action
80	Cron	glite cron jobs have execute permissions	#15638	CLOSED		glite-security-config version 1.2.4 glite-WMS version 2.3.3 glite-iperf-1.3.2-1.i386.rpm glite-udpmon-1.3.10-i386.rpm, checked by Di
81	Yaim	FTS needs configuration scripts	#16149	CLOSED		glite-yaim-3.0.0-6.noarch.rpm
82	NW Server	NW Server "Authentication Problem"	#15761	CLOSED		Due to #15638, checked by Di
83	Yaim	dcap should be gsidcap in yaim. New function submit to test.	#16161	CLOSED		glite-yaim-3.0.0-6.noarch.rpm, checked by Di
84	Yaim	DPM/LFC misconfiguration of the /etc/shift.conf	#16162	CLOSED		RC4, checked by Di

Issue 67 * may be back.

Found in Pre-production

Number	Component	Description	Savannah	Status	State	Action
76	UI/WN	A relocatable UI/WN is needed	15998	Requires fixes to glite configuration	REOPENED	Robert
85	gLite RB	Logmonitor crashed	#16191	Occurs when people do nasty things while debugging	CLOSED	None
87	Yaim	Bug in path setup of grid pool accounts	#16319		CLSOED	glite-config-1.6.29-0 package, checked by Di
88	WMS	Configuration for bulk submission needed	#16294		CLSOED	glite-yaim-3.0.0-9, checked by Di
89	Yaim	glite-yaim-3.0.0-5: 'gliteCE' parameter missing in the example file	#16304	Wiki error, parameter removed	CLOSED	None
90	Yaim	YAIM install_node script fails to recognise lcg-CE_torque node type	#16347	Patch available	FIXED	Fixed in Yaim head
91	Yaim	"Connection refused" during YAIM configuration of Torque	#16339		OPEN	Di
92	glite CE	Software TAG publish	#10872	Fixed in RC3	CLOSED	checked by Di
93	BDII	site does not publish GlueSiteUniqueID	#16248	Won't fix	CLOSED	NONE
94	WMS	WMProxyEndpoints attribute empty in glite_wms.conf	#16330	duplicated with the bug no #16294.	CLOSED	checked by Di
95	Yaim	glite-yaim does set queue.acl.groups	#16220	Fixed in YAIM head	FIXED
99	gliteCE	APEL	#16425	Can not fix before the release	Closed	None
101	config_gip	GlueHostOperatingSystemVersion hardcoded	#16425		FIXED	Fixed in Yaim head

Issue 67 is being encountered again in PPS.

Found by ROC Testers

Number	Component	Description	Savannah	Status	State	Action
86	dCache	Reset the databases	#16388	Patch available for YAIM, needs adding	CLOSED	glite-yaim-3.0.0-9, checked by Di
96	Install	Wrong versions for lcg-info-dynamic-scheduler	#16355	To be done for RC4	CLOSED	RC4, checked by Di

Carried forward

Number	Component	Description	Savannah	Status	State	Action
50	L&B	status not updated for nodes of a large collection Apr/6 - bugs identified causing high CPU usage of glite-proxy-renewd, for this new bug was raised 16050, we'll track both here Apr/10 - file descriptor limit can be increased, but Francesco has a proper fix. Will produce the tag to be tested on CTB.	15189 16050	Working with ~400 jobs, still problems with 1000. Could this be related to slow matchmaking or a limited number of available CEs? More work	FIXED	Included in integration build for 3.0
57.1	WMS	Job aborts after proxy renewal Apr/4 - new bug 15905 Apr/5 - nonews Apr/10 - waiting for DI's testing results (on his private setup, not CTB). In any case Francesco reports problem has been seen in dev. Apr/12 - the tag with fix for #15905 arrived in integration. To be included in RC3.	15905(15643) (13963)	Fixed in RC3	CLOSED	checked by Di
69	gliteCE	How should we manage publishing of SoftwareRunTimeEnvironment Apr/6 - add gridftp server to gLite CE, configure exactly as LCG CE, assigned to Louis Apr/10 - change being prepared for RC3 Apr/12 - Robert is working on it.	16159	On the gLite 3.0 timescale we will have to put gridFTP on the gliteCE. We must make sure, if possible, that we configure to restrict access only to sgms, and to allow them only to edit a single file.	CLOSED	glite-CE-2.4.5-2, checked by Di
72	NS	NS produces authentication errors after 3-4 days Apr/6 - nothing heard from developer, Francesco will ping him Apr/10 - A similar problem has been observed on RC2, not clear if identical, to be analysed on PPS.	15761	Developer investigating node in this state. Is this time or activity related? Due to #15638	CLOSED	checked by Di
73	Doc	The port table has to be updated for all services in the release Apr/6 - nothing obtained, Zdenek will ping John White Apr/10 - John is contacting developers to supply their parts, doc in preparation. Apr/12 - list for gLite delivered. Needs to be merged with LCG list. John working on it.		Ian N contacted	OPEN	Romain John Ian N

Issues with fixed and CLOSED

Number	Component	Description	Savannah	Status	State	Action
24	FTS	Provide FTS server YAIM component Apr/6 - done by Gavin, not in CTB repository yet Apr/10 - moved to CLOSED		FTS node still requires some manual configuration but is mostly yaim now. FTA has a full yaim component, to be tested	CLOSED	Gavin.
24.1	FTA	Provide FTA YAIM component		Ready for yaim 3.0.0-3. FTA can be installed on the FTS server or on a separate machine	CLOSED	Gavin
34.1	DPM/LFC	Race condition in srm v2 methods		Fixed in 1.5.5	CLOSED	In cert
34.2	DPM/LFC	Python LFC interface broken in 1.5.3, should regress to 1.5.2 python interface.		Fixed in 1.5.5 (without reverting to old interface)	CLOSED	1.5.5 in cert
34.3	DPM/LFC	bug fix requiring db update		Fix committed. update integrated into upgrade script	CLOSED	1.5.5 1.5.5 in cert
34.4	DPM/LFC	Fresh install problems with virtual id mappings		Fixed in yaim config	CLOSED	Jean-Philippe/Louis
34.5	DPM/LFC	rfio problem preventing correct functioning with more than one pool node.		Fixed in 1.5.5	CLOSED	1.5.5 in cert
34.6	DPM/LFC	Oracle error due to locks for virtual ids		Fixed in 1.5.5	CLOSED	1.5.5 in cert
34.7	DPM/LFC	DPM upgrade script doesn't work with more than one VO			CLOSED	Sophie, fixed
36.1	LB	With >5000 jobs the interlogger is in a state where events are not forwarded.	15217	A possible contributor to 15050	CLOSED	Fix in the repository. Daniele confirms fix
39	DPM/LFC	To be statically linked against MySQL 4.1	15421	1.5.5 linked against MySQL 4.1	CLOSED	1.5.5 in cert
40	lcg-RB	RB needs modifications to run with new Condor, both a config and updates to some binaries eg condor_gridmanager	15417	Fixed in yaim	CLOSED	Ready for cert
41		The ./glite-lfc-client-config-config.py script cannot be found	15425		CLOSED	Issue in glite-wn.cfg.xml template, ready for test
42	WMS/gridFTP	WMS gridFTP crashing with > ~56 CAs	15383	New tag produced 8th, to build 10th	CLOSED	fix confirmed
43	gliteCE	gliteCE pointing to separate torque server	15424		CLOSED	Robert - fixed in yaim, ready for cert
44	gliteCE	In a glite 3.0 CE(glite flavour) the LSF gip script points to a non existing file	15434		CLOSED	Fixed in CE config script
45	gliteCE	It should be documented that the logparser daemon MUST be started on a glite CE	15432		CLOSED
46	WMS	Job submission from a glite 3.0 WMS (glite flavour) to a glite 3.0 CE (glite flavour) fails	15426		CLOSED	problem with a missing rpm, now it's a dependency
47	VOBOX	unable to install glite-UI untop of glite-VOBOX	15411		CLOSED	glite-VOBOX packate to incorporate all of glite-UI
48	FTS	FTS - YAIM configuration: "error reading information on service rgma-gin","... bdii"	15410		CLOSED	rpm missing, fix ready for cert
49 *	UI	remove links to edg-* commands	15330	Update 28/03/2006; links to be removed	CLOSED	Fabrizio
51	WMS	glite-wms-wm crashes under some unknown bdii or ism condition	15098		CLOSED
52	WMS	submission of collection always fails	15095		CLOSED
53	WMS	WMProxy not correctly configured	14861		CLOSED
54	WMS	ISM keeps CEs removed from a BDII	14930		CLOSED
55	yaim/CE	It seems it is not possible with YAIM to map the VOs to a unique queue when using LSF	15520		CLOSED	invalid; site-info.def misconfiguration
56	combined UI	No dpm client in the combined UI	15514		CLOSED	Joachim - fix ready
57 *	WMS	VOMS proxy renewal failing on WMS	15643 (13963)	Problem identified in handling of vomses file in voms library. Fix on CVS, to tag	CLOSED	Di, Mario, checked by Di
57.1	VOMS	VOMS proxies with '=NULL' format for attributes must be supported	-		CLOSED	Maarten, checked by Di
57.2	WMS	Apr/5 - New critical bug will be posted by Francesco for VOMS authentification, not observed yet because DPM hasn't been used with ACL's so far. Problem discovered as by-product of analysis of pt.57.1 above. Apr/6 - still in discussion, no bug raised yet, no problem for GRAM based jobs Markus: contacted JPB for a test of setacl from the WN, Gilbert doesn't have one, current DPM tests cannot be adapted, write a new one (who? Gilbert? JPB? Markus to solve) Apr/10 - inexistent problem! After much deeper analysis the original worries appeared unfounded. Closing.	??	Francesco to open the bug	CLOSED	Francesco
58	gliteCE	lcmaps/lcmaps.db updated for VOMS roles	-	fix in CVS	CLOSED	Alberto - in integration repository
59	installation	renamed glite meta rpms		In integration repo	CLOSED	Joachim/Oliver
60	WMS	AccessControlbaseRule & LSF	15642	CLOSED
62	gliteCE	Blah submission from a glite 3.0 CE (glite flavour) to an LSF queue does not work Apr/6 - in the RC2 repository	15674	Two solutions i) run log parser on head node ii) copy log file locally onto a non-AFS partition. Alessandro to investigate (ii). (i) difficult because parser is not available as a standalone rpm (for 3.1). Job submission still failing.	CLOSED	Alessandro
63	gliteCE	Jobs remaining in ready state despite finishing Apr/5 - We may need to upgrade Condor to 6.7.16 - no code change 6.7.18 - requires one-line change, would be submitted by Francesco, better Condor support expected if higher version used. -> Francesco to check existing bad-state-gliteCE's for a new hint Apr/6 - bad state gliteCE's revealed that cron has been auto-updated, now requires crotab not to have exec permission for security, but all(?) our crontabs are created with that perm bit on(!!), CTB has to now verify all nodes and remove exec perm by hand for RC2, YAIM script needs to be changed, but some crontabs are hardcoded, to cleanup for after-RC2. After-RC2 testing of condor 6.7.16 may be the best, 6.7.18 has new problems, better wait for 6.7.19 in a few weeks, Francesco thinks. This may be a long problem before we get to the right solution Apr/10 - the bug will be raised against YAIM install to fix the crontab exec bit problem, which seems to have caused all seen problems (after increased timeout has fixed some, the original assessment that it didn't is considered incorrect now) Problem is being closed now waiting for new observations from the PPS, will be reopened if appropriate after PPS assessment. Francesco will continue testing condor 6.7.19 just in case.	15688	Adjustement of timeouts suggested by Francesco has not worked. This could be related to problems in the interaction of LCG LB clients and the glite LB, which will always happen in the gliteWMS->LCG-2_7_0 submission chain.	CLOSED	Alessandro Francesco
64	gliteCE	after restart of a gliteCE jobs fail for a brief period Apr/5 - moved to CLOSED	15685	Issue to be refered to the Condor group. Anticipated effect in production is minimal	CLOSED	condor
65 *	WMS	gLite 3.0 job wrapper has bad kill usage	15710	in integration repo	CLOSED
66	savannah	Migration of existing issues to new Savannah project Apr/6 - all bugs needed tracking for gLite 3.0 were moved		categories created, users and issues not yet transfered	CLOSED	Nick
67 *	WMS/UI	glite-job-list-match fails, undefined symbol	15724	Retag unchanged code, rebuild and see if problem persists; tag sent 29/03/2006	CLOSED
68	inst	It is sometimes necessary to remove boost-g3 before installing glite3.0 Apr/5 - must be fixed, who can fix it? -> must be put in Savanah as a bug, Joachim will submit the bug Apr/6 - bug submitted, temporary solution by Louis with new rpm Apr/10 - though being able to keep only one boost-g3 would be the correct solution, given the fact it needs code changes forces us to stay with the Louis' temporary solution. Closed.	15986	We need to find out if we can remove the 'provides boost' from boost-g3. The rpm has Build Host: pc-cel3-build Problem must be tracked until we have a proper solution or a full confidence in the current one	CLOSED	Joachim Louis
70	WMS	large job collection submission and cancel through WMproxy didn't work Apr/5 - moved to CLOSED	15769	17MB MySQL limit implemented, Mario confirms issue fixed	CLOSED
71	WMS	fix for speeding up match-making for bulk submissions of a factor of 5 (from ~10 to ~2 s per job) Apr/5 - moved to CLOSED	15806	This comprises two fixes, first of which relates to brokerinfo files, which will be ready to tag by 31/03/2006. The other component to the fix will not be ready on this timescale, but is not tied to the first.	CLOSED
74	FTS	MySQL FTS has wrong db schema version Apr/6 - update of operating procedures was needed	15875	Gav fixing	CLOSED	Gav
74 *	UI	bad distinction between file/directory of $GLITE_LOCATION/etc/vomses	15734	This halts the UI config	CLOSED
75	FTS	Every so often a "Failed to get proxy certificate" error is obtained when submitting a transfer Apr/5 - moved to CLOSED	15866	Fixed in yaim due for RC2 release	CLOSED	Gav
77		Apr/5 - gLite CE won't work with LSF queues lsf_submit.sh must not source ~/.login Apr/5 - The tag with the QF for this bug is: glite-ce_R_1_5_12 Apr/6 - tag is in the RC2 repository, ready to test	15985		CLOSED
78	FTS	Job status is Failed after four re-tries, but file was copied to the destination. Apr/6 - closed by Gavin as not reproducible.	15987	The wrong RPMs were tagged for the original gLite 3.0 build. Need to check this in RC2	CLOSED	Gavin
79	VOMS	Apr/5 - ACs list contains wrong group/role values Apr/6 - needs to be tested by Di	15692		CLOSED	Maria Di