Present:
	Glasgow: Sam
	Liverpool: Jon, Stephen, Mark
	Oxford: Ewan, Pete
	Imperial: Duncan
	QMUL: Chris
	Sheffield: Elena
	Manchester: Alessandra
	RAL: Brian, Andrew, Jens



[09:50:06] Sam Skipsey joined
[09:58:47] John Bland joined
[09:59:52] Andrew Sansum joined
[10:01:05] Stephen Jones joined
[10:01:06] John Bland mic's playing up
[10:01:07] Mark Norman joined
[10:01:11] Ewan Mac Mahon joined
[10:01:12] John Bland we had an email last night asking for resources
[10:01:14] Duncan Rand joined
[10:01:33] John Bland just a few TB for now, they're still deciding on exactly what they want
[10:01:44] Christopher Walker joined
[10:02:41] Elena Korolkova joined
[10:02:47] Alessandra Forti joined
[10:03:02] Brian Davies joined
[10:03:14] Pete Gronbech joined
[10:05:48] Andrew Sansum left
[10:06:28] Govind Songara joined
[10:06:56] Andrew Sansum joined
[10:07:38] Elena Korolkova The question is do we really need to upgrde from 1.0 to 1.8 if we will be ask to move to EMI in couple of month
[10:07:59] Elena Korolkova sorry from 1.8.0 to 1.8.2
[10:10:57] Matthew Doidge joined
[10:11:19] Alessandra Forti 1.8.2 should be in EMI by now shouldn't it?
[10:11:59] Andrew Sansum left
[10:13:11] Elena Korolkova There was email from Maarten last week in which it was not recommended to switch to EMI dpm ATM. 
[10:14:11] Sam Skipsey 1.8.2 *is* in EMI by now. IIRC, the issue with the EMI dpm (which is actually an issue with EPEL, and with Globus) is actually fixed. (In any case, it will be properly fixed in the 1.8.3 release which is due in about a week or so.)
[10:14:12] Elena Korolkova And there is no much experience in movimg to EMI 1.8.2.
[10:14:59] Andrew Sansum joined
[10:15:20] Elena Korolkova I don't want to switch from stable storage to unstable one just to tick the box
[10:17:00] Sam Skipsey Right, but UMD is the "stable" release (or at least, the tested release) version of EMI.
[10:17:23] Andrew Sansum left
[10:17:35] Duncan Rand but it contains the latest version fixing significant bugs or introducing important features.
[10:17:41] Elena Korolkova Does somene have UMD 1.8.2 in UK?
[10:17:46] Duncan Rand https://twiki.cern.ch/twiki/bin/view/LCG/WLCGBaselineVersions
[10:17:50] Elena Korolkova installed
[10:17:51] Sam Skipsey and Durham have been running a relatively stable EMI DPM for a while and it seems no worse than the gLite one.
[10:18:21] Elena Korolkova Durham is not good example
[10:19:16] Sam Skipsey Why, Elena?
[10:19:41] Elena Korolkova it's not a stable site




[09:50:06] Sam Skipsey joined
[09:58:47] John Bland joined
[09:59:52] Andrew Sansum joined
[10:01:05] Stephen Jones joined
[10:01:06] John Bland mic's playing up
[10:01:07] Mark Norman joined
[10:01:11] Ewan Mac Mahon joined
[10:01:12] John Bland we had an email last night asking for resources
[10:01:14] Duncan Rand joined
[10:01:33] John Bland just a few TB for now, they're still deciding on exactly what they want
[10:01:44] Christopher Walker joined
[10:02:41] Elena Korolkova joined
[10:02:47] Alessandra Forti joined
[10:03:02] Brian Davies joined
[10:03:14] Pete Gronbech joined
[10:05:48] Andrew Sansum left
[10:06:28] Govind Songara joined
[10:06:56] Andrew Sansum joined
[10:07:38] Elena Korolkova The question is do we really need to upgrde from 1.0 to 1.8 if we will be ask to move to EMI in couple of month
[10:07:59] Elena Korolkova sorry from 1.8.0 to 1.8.2
[10:10:57] Matthew Doidge joined
[10:11:19] Alessandra Forti 1.8.2 should be in EMI by now shouldn't it?
[10:11:59] Andrew Sansum left
[10:13:11] Elena Korolkova There was email from Maarten last week in which it was not recommended to switch to EMI dpm ATM. 
[10:14:11] Sam Skipsey 1.8.2 *is* in EMI by now. IIRC, the issue with the EMI dpm (which is actually an issue with EPEL, and with Globus) is actually fixed. (In any case, it will be properly fixed in the 1.8.3 release which is due in about a week or so.)
[10:14:12] Elena Korolkova And there is no much experience in movimg to EMI 1.8.2.
[10:14:59] Andrew Sansum joined
[10:15:20] Elena Korolkova I don't want to switch from stable storage to unstable one just to tick the box
[10:17:00] Sam Skipsey Right, but UMD is the "stable" release (or at least, the tested release) version of EMI.
[10:17:23] Andrew Sansum left
[10:17:35] Duncan Rand but it contains the latest version fixing significant bugs or introducing important features.
[10:17:41] Elena Korolkova Does somene have UMD 1.8.2 in UK?
[10:17:46] Duncan Rand https://twiki.cern.ch/twiki/bin/view/LCG/WLCGBaselineVersions
[10:17:50] Elena Korolkova installed
[10:17:51] Sam Skipsey and Durham have been running a relatively stable EMI DPM for a while and it seems no worse than the gLite one.
[10:18:21] Elena Korolkova Durham is not good example
[10:19:16] Sam Skipsey Why, Elena?
[10:19:41] Elena Korolkova it's not a stable site
[10:20:09] Alessandra Forti it's also a small site not running analysis
[10:20:13] Alessandra Forti in atlas
[10:20:29] Sam Skipsey Sure, but what kind of "instability" are you worrying about?
[10:20:50] Alessandra Forti I'm not worried I'm on 1.8.2  
[10:21:02] Alessandra Forti but I can understand elena objections
[10:21:03] Sam Skipsey Remember there's *no functional difference* between the EMI and gLite packagings of DPM (other than the repositories they're in and the places they install things)
[10:21:49] Sam Skipsey So, if you have worries about "stability" they should be with 1.8.2, which as Alessandra has demonstrated is installed at a very heavily used site in the UK.
[10:21:59] Alessandra Forti these are not minor changes.
[10:22:11] Elena Korolkova I remember that Alessandra was trying to install EMI headnode but finally gave up and installed glite version
[10:22:22] Alessandra Forti I never tried EMI
[10:22:31] Sam Skipsey You're thinking of Ewan, Elena.
[10:22:43] Sam Skipsey And we're pretty sure we understand where the issue was there.
[10:22:47] Alessandra Forti when I installed 1.8.2 was only in glite3.2
[10:22:56] Alessandra Forti I haven't had any problem with that
[10:23:22] Sam Skipsey and *my* personal experience is that installing EMI (or migrating to EMI) is precisely as hard as installing gLite3.2
[10:24:24] Santanu Das joined
[10:24:31] Sam Skipsey I'm not saying everyone should rush to EMI installs - I'm perfectly happy with Alessandra being on 1.8.2 glite - but I don't think we should react with fear and horror at the very concept.
[10:25:06] Ewan Mac Mahon I have every intention of jumping to EMI as soon as I can get my upgrade going, and the last bit of hardware should arrive tomorrow.
[10:25:30] Alessandra Forti I think it is understandable that a very stable site doesn't want to upgrade just for the sake of it.
[10:26:23] Santanu Das I upgraded to UMD; working good so far.....
[10:26:29] Sam Skipsey Sure, but ultimately you're delaying the inevitable.
[10:26:58] Elena Korolkova if I have EMI head node should I immeadiately move dpm servers to EMI?
[10:27:03] Sam Skipsey No.
[10:27:15] Sam Skipsey As I mentioned, EMI and gLite releases are functionally identical.
[10:27:32] Sam Skipsey So a gLite head can happily manage EMI servers, and vice versa.
[10:27:43] Ewan Mac Mahon I'm just having a look at the GridPP28 schedule and I'm not seeing much slack space, unless we want some people to go a day early
[10:27:52] Ewan Mac Mahon and have a storage meeting at the same time as the PMB is on.
[10:28:02] Santanu Das As Sam said: all my disk-servers are running gLite, head node is UMD
[10:28:14] Elena Korolkova I need a spare machine for that. We expecting it to arrive this month
[10:28:54] Duncan Rand sounds like you need HDFS
[10:29:01] Sam Skipsey That's fine. As I said, I'm certainly not pushing everyone to jump to EMI right now. I'm just asking for people to consider that they'll have to move at some point, and it's better if they at least consider their move now.
[10:34:17] Govind Songara Any idea when EMI DPM for SL6 set to released..?
[10:35:28] Sam Skipsey Not completely. Although 1.8.3 is being tested on SL6...
[10:35:46] Elena Korolkova Wednesday 18 April afternoon?
[10:36:16] Elena Korolkova at GridPP 28th?
[10:37:55] Ewan Mac Mahon I suspect we'd need several tests in a suite, 
[10:38:07] Christopher Walker Non LHC VO support - and a data model for them would be useful too. 
[10:38:26] Ewan Mac Mahon (a bit like old CPU benchmarks having separate integer and FP scores)
[10:40:33] John Bland left
[10:40:34] Elena Korolkova left
[10:40:34] Govind Songara left