[prev in list] [next in list] [prev in thread] [next in thread] 

List:       npaci-rocks-discussion
Subject:    [Rocks-Discuss] Re: Some (but not all) nodes stalling upon reinstall
From:       Philip Michael Papadopoulos <ppapadop () uci ! edu>
Date:       2020-04-14 22:01:28
Message-ID: CAF2yGiX9ehgVkvOgY8+vPd19=0bX6GCTDc0oda-rw_PdLEOgQg () mail ! gmail ! com
[Download RAW message or body]

robert, You can attach the output of both as files and send directly to me.
I'll take a look.

-P


On Tue, Apr 14, 2020 at 2:24 PM Rovetti, Robert J. <Robert.Rovetti@lmu.edu>
wrote:

> Thanks so much Phillip.
> 
> My IT guy might not be able to access the server room and inspect the
> switch until later (due to campus lockdown), hopefully sooner rather than
> later.
> 
> As for the output of "rocks list host profile":    It runs without error
> on both nodes (node with successful update, and node without successful
> update).  However, the output is quite different between the two; one seems
> to be a highly scrambled / reordered version of the other, so it is quite
> difficult to compare them directly.  Perhaps that is a consequence of one
> of the nodes updating successfully.    Is there anything in the profile
> output that I should look for?
> 
> Thanks,
> Robert
> 
> 
> -----Original Message-----
> From: npaci-rocks-discussion-bounces@sdsc.edu <
> npaci-rocks-discussion-bounces@sdsc.edu> On Behalf Of Philip Michael
> Papadopoulos
> Sent: Saturday, April 11, 2020 8:38 AM
> To: Discussion of Rocks Clusters <npaci-rocks-discussion@sdsc.edu>
> Subject: [Rocks-Discuss] Re: Some (but not all) nodes stalling upon
> reinstall
> 
> what kind of switch do you have?
> It's possible that some ports on the switch are configured differently
> than others.
> 
> When a node PXE boots, the network interface goes up and down several
> times (This isn't peculiar to Rocks, it's the way native installation
> works).
> 
> Some switches see this as "flapping" and disable the port for a period of
> time. Long enough for nodes to "timeout"
> 
> Also,
> on the frontend
> 
> # rocks list host profile <node that works> and # rocks list host profile
> <node that doesn't work>
> 
> Are there any errors in either case - there shouldn't be.  Do  the outputs
> look similar? They should modulo host names and other host-specific items.
> 
> -P
> -------------- next part --------------
> An HTML attachment was scrubbed...
> URL:
> http://lists.sdsc.edu/pipermail/npaci-rocks-discussion/attachments/20200411/27dd39a5/attachment.html
>  
> 

-- 
Philip Papadopoulos, Ph.D
Director, Research Cyber Infrastructure Center
University of California, Irvine
E-mail: ppapadop@uci.edu
Phone: (949) 824-5343
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.sdsc.edu/pipermail/npaci-rocks-discussion/attachments/20200414/9783096b/attachment.html \



[prev in list] [next in list] [prev in thread] [next in thread]