home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!pipex!unipalm!uknet!mcsun!dxcern!news
- From: clusmgr@vxcern.decnet.cern.ch (LETERTRE Christiane)
- Newsgroups: comp.os.vms
- Subject: Re: VMS 5.5-1 / MIVc, satellite prevents the boot hos
- Message-ID: <1992Nov5.170152.13650@dxcern.cern.ch>
- Date: 5 Nov 92 17:01:52 GMT
- References: <1992Nov4.122801.3938@arizona.edu> <1992Nov4.164802.14651@dxcern.cern.ch> <921105000706.000003E0@MARVIN.CTSS.CO.UK>
- Sender: news@dxcern.cern.ch (USENET News System)
- Reply-To: LETERTRE@VXCERN.DECNET.CERN.CH
- Organization: CERN
- Lines: 73
-
-
- In article <921105000706.000003E0@MARVIN.CTSS.CO.UK>, VERKADE@CTSS.CO.UK (Herman Verkade) writes:
- |>Path: dxcern!mcsun!uknet!ctssuk!VERKADE
- |>From: VERKADE@CTSS.CO.UK (Herman Verkade)
- |>Newsgroups: comp.os.vms
- |>Subject: Re: VMS 5.5-1 / MIVc, satellite prevents the boot hos
- |>Message-ID: <921105000706.000003E0@MARVIN.CTSS.CO.UK>
- |>Date: 5 Nov 92 00:07:06 GMT
- |>References: <1992Nov4.122801.3938@arizona.edu> <1992Nov4.164802.14651@dxcern.cern.ch>
- |>Organization: CompuThoughts Software Solutions (UK) Ltd.
- |>Lines: 37
- |>
- |>In article <1992Nov4.122801.3938@arizona.edu>, leonard@telcom.arizona.edu (Aaron Leonard) writes:
- |>>In article <1992Nov4.164802.14651@dxcern.cern.ch>, clusmgr@vxcern.decnet.cern.ch () writes:
- |>>|
- |>>| We recently upgraded one of our VAX clusters from VMS 5.4-1A to VMS 5.5-1.
- |>>| This cluster was already an MIVc but without any satellite.
- |>>| Just after the upgrade, we introduced one satellite in the cluster.
- |>>|
- |>>| Since this date, if the satellite is UP, we are unable to reboot the boot host.
- |>>| Each time the reboot is hung, very early in the startup procedure (see below).
- |>>| We need to kill the satellite to complete the reboot.
- |>>|
- |>.
- |>.
- |>>|
- |>>| Sometimes the hang occurs a little later, after having mounted one or two disks
- |>>| mounted by SYLOGICALS. We have no problem to reboot the satellite after the
- |>>| boot host has been rebooted.
- |>>|
- |>>| Is this a known problem with MIVc and VMS 5.5-1 or did we make something wrong?
- |>>| Any help will be greatly appreciated as we plan to upgrade another MIVc next
- |>>| week end.
- |>>
- |>>We have seen similar behavior in a similar configuration (LAVc, VMS V5.5-1).
- |>>My hunch, though I haven't been able to prove it, is that there is a bug in the
- |>>filesystem such that an MSCP server hangs in VMOUNT when attempting to mount
- |>>a disk if there is some kind of pending I/O from a MSCP client.
- |>>
- |>>Has ANYONE managed to reboot a boot server under VMS V5.5-1?
- |>
- |>Yes, I have seen it done, but I have also seen a site experiencing the above
- |>problems. The only `odity' there was that the satellites booted off another
- |>system disk than the boot member (In fact, the boot member ran V5.4-2 and the
- |>satellites ran V5.5-1). This problem was 100% reproduceable, but TSC of course
- |>suggested to upgrade the boot member to V5.5-1 before they would look at it.
- |>Does this resemble your configurations?
- |>
- |>Herman
- |>
- --
- No, this does not really resemble our configuration: we have only one system
- disk (shadowed) and we are already running VMS 5.5-1.
- The only oddity with this cluster is that the satellite (VSOPS) is not in the
- same DECNET area as the 2 other nodes, but I found nodo documentation which
- forbids such a configuration.
- Anyway, this morning we decide to get away of this peculiarity! So, using
- CLUSTER_CONFIG, I removed the satellite, then added it with the new DECNET
- number. Unfortunately, the satellite did not reboot after its reintroduction
- in the cluster.
- The boot failed with "No connection to disk server VXCRNA", just after
- displaying the good line "System loaded from node VXCRNA".
- We are more and more in the dark, and any suggestion will be very well
- appreciated. Thanks in advance.
- ------------------------------------------------------------------------------
- Christiane LETERTRE
-
- CERN / European Laboratory for Particle Physics
- CH-1211 GENEVA 23
- Tel: (022) 767 28 54
-
- Mail address: LETERTRE@VXCERN.DECNET.CERN.CH
- ------------------------------------------------------------------------------
-