[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference tape::infoserver

Title:InfoServer (Ethernet System Server)
Notice:Much more than just a CD Servere TAPE::
Moderator:TAPE::STONEHAM
Created:Wed Mar 21 1990
Last Modified:Thu Jun 05 1997
Last Successful Update:Fri Jun 06 1997
Number of topics:2223
Total number of notes:9773

2194.0. "Device Timeout on Infoserver Services" by CSC32::S_WAHL () Wed Feb 12 1997 01:12

Customer is running 3.2 kernel software on an Infoserver 150 and VAX 6.0.  
Intermittently the customer is experiencing device timeouts on a cd services 
offered by the infoserver.  The strange thing about it is its always the same
reader that is having the problem.  There are seven readers in the cabinet.  
She has replaced this device with two other readers and the same problems 
occur.  Customer has already had HW (outside vendor) check out the 
configuration.  Apparently they say nothing is wrong.  

I still believe there is a HW problem.  The customer is unable to get this 
device back at all.  It goes into mount verification and then eventually 
times out.  There is no consistency in the time the device takes to do this.

Could this be an exhausted resource?  Why are none of the other readers 
failing if this was the case?  Any and all help would be appreciated here.
I could not really find anything in STARs that pointed to a problem like
this or a recommended patch.


Thanks,

Susan M. Wahl

T.RTitleUserPersonal
Name
DateLines
2194.1This info might helpWAYLAY::GORDONResident Lightning DesignerWed Feb 12 1997 12:1917
	Well, let's start from the beginning...


	What device type is failing?

	Is it always the same CD failing?

	Is this device the last one on the bus?  How long are the cables?
	Who made the cabinet?

	What does the undocumented SHOW POOL command have to offer in
terms of resource exhaustion.

	(unlikely to help but) can you get them to upgrade to V3.4


						--Doug
2194.2more detailsCSC32::S_WAHLWed Feb 12 1997 15:0335
>>>	What device type is failing?

	RRD42, replaced w/another RRD42

>>>	Is it always the same CD failing?

	Yes.

>>>	Is this device the last one on the bus?  How long are the cables?
	Who made the cabinet?

	Chasis has six devices that are daisyed.  It is the third device.
	I think I understand - there is a Hard Drive, then two addl readers,
	then this chasis.

>>>	What does the undocumented SHOW POOL command have to offer in
>>>	terms of resource exhaustion.

$ sh mem/pool
              System Memory Resources on 12-FEB-1997 09:57:26.60

Dynamic Memory Usage (bytes):      Total        Free      In Use     Largest
  Nonpaged Dynamic Memory       10735616	7892096	  2843520    6898240	
  Paged Dynamic Memory           6376448	4268864	  2107584    4241392

    Does this help?
    
    Thanks,
    
    Susan
		
>>>	(unlikely to help but) can you get them to upgrade to V3.4

	Customer is willing to update.

2194.3WAYLAY::GORDONResident Lightning DesignerWed Feb 12 1997 16:1943
>>>	Is it always the same CD failing?

>	Yes.

	If they move the CD to another drive, does the failure follow the CD?


>	Chasis has six devices that are daisyed.  It is the third device.
>	I think I understand - there is a Hard Drive, then two addl readers,
>	then this chasis.

	Eh?  It's a 150, so it's either on the A or B bus.  I'm guessing
that it's the A bus from what you said, so the hard drive and one CD drive
are in the InfoServer cab.  What's the external cab and who made it?  Is
it a bunch of tabletop RRD42s stacked up?  How long are all the cables
between items?


	I meant SHOW POOL on the InfoServer.  Should produce a display like:

Gumby> sh pool

Digital Equipment Corporation
InfoServer 1000 V3.3         (BL26)       System Image:  Jun 26 1995 17:50:05

Server Name:                  GUMBY       Uptime:                060 23:11:53

Pool Size:                   424152       Cache Size:                 3164456
Pool Available:              294688       Pool poisoning is enabled.
% of Pool Free:                 69%

Element  Total      Allocated Allocated   Free         Free         Allocation
Size     Elements   Elements  Bytes       Elements     Bytes        Failures

  0-104     20         17         1008        3             192         0
105-288     40         38         8168        2             400         0
289-1016    37         36        16960        1             520         0
1017+       56         54       102104        2          293576         0

Totals:    153        145       128240        8          294688         0

Smallest free element size in bytes:        48
Largest free element size in bytes:     289480
2194.4Further detailsCSC32::S_WAHLThu Feb 13 1997 20:4538
    Doug,
    
    The problem does not seem to follow the cd.  The CD will work in any
    other RRD42.  The chasis is made by Line America model - #541.  Cable
    length is 1-2 feet.  Customer repeated not very long at all.
    
    The SHO POOL command from the Infoserver says:
    
    Infoserver 150 V3.2(BL22C)			SYSTEM IMAGE: 30-SEP-1994
    Server Name= ATHENA				Uptime = 0 14 7:25:42
    Poolsize: 439512				Cachhe Size: 3164456
    Pool Avail: 302232				Pool Poisoning: Disabled
    % of Pool Free - 68% 
    
    Element	Total	Allocated	Allocated    Free    Free   Alloc	
    Size	Element	Elements	Bytes        Elem    Bytes  Fail
    0-104	  28	   23 		1208           5      248     0	
    105-288       75	   75           17584          0	0     0
    289-1016	  	   68		30872          1      680     0  
    1017+	  52       48		85840          4    301288    0
    
    Totals
    
    		 224	  214	       135504	      10    302212    0
    
    Smallest Free Elements Size in Bytes = 24
    Largest Free Elements Size in Bytes = 286656
    
    I typed the above information as read by the customer.  The value that
    is missing under the Total Element column was deleted by me - it was 
    not origianlly blank.  The customer was not available when I called her
    back for the data.
    
    Hope this helps.
    
    Thanks,
    
    Susan
2194.5I'm down to some pretty far out speculations in cases...WAYLAY::GORDONResident Lightning DesignerFri Feb 14 1997 12:2035
	OK - it doesn't look like resource problems on the InfoServer. (The
missing value is of no importance, the last column is what's really 
significant in this case.)

	I'd suspect one of the following things:

	a) SCSI Termination in the external box.

	b) bus length.  Remember the bus is a total of all the internal cable
	   in the InfoServer, plus the connecting cable, plus the cabling
	   in the external cab.

	c) flakey cable. Unlikely, but possible.

	d) Termpower problems.  This one is tough and based on what I just
	   looked up, probably not likely.



	When this happens, what does SHOW DEVICE say about the drive in
question.  Most obvious thing to look for is if it has "dismounted" the
CD (no volume label visible.)  If so, the InfoServer automount poll is getting
an error and thinks the CD has been ejected.  Of course in that case it should
run down the aautomounted service too.

	One other possibility.  Is there any other InfoServer offering the
same service name, especially if the service name is created and not 
automounted?  If so the client could be failing over to a service name that
isn't really what it expects.  Again, unlikely but I wracking my brain for
scenarios that could fit.

	There's not much more help I can offer.  I'm no longer even with
InfoServer Engineering.

					--Doug