SFTP SSH_ALLOC_ERROR during initial connection

[Allan Large]

HP TCP/IP Services for OpenVMS Alpha Version V5.4 - ECO 7 on a AlphaServer DS25 running OpenVMS V7.3-2

$ prod sho hist
----------------------------------- ----------- ----------- --------------------
PRODUCT KIT TYPE OPERATION DATE AND TIME
----------------------------------- ----------- ----------- --------------------
DEC AXPVMS TCPIP V5.4-15ECO7 Full LP Install 13-FEB-2008 11:30:38
DEC AXPVMS TCPIP V5.4-15 Full LP Remove 13-FEB-2008 11:30:38
DEC AXPVMS TCPIP_ECO V5.4-156 Patch Remove 13-FEB-2008 11:30:38
DEC AXPVMS TCPIP_ECO V5.4-155 Patch Remove 13-FEB-2008 11:30:38
DEC AXPVMS TCPIP_ECO V5.4-154 Patch Remove 13-FEB-2008 11:30:38
DEC AXPVMS TCPIP_ECO V5.4-151 Patch Remove 13-FEB-2008 11:30:38


SFTP / SSH was working fine on this system since Feb 2008. After a system reboot due to power failure a month ago, when trying to SFTP into the system the server side TCPIP$SSH_RUN.LOG shows the error

%TCPIP-F-SSH_ALLOC_ERROR, ssh memory allocation error

On the client side it is reported as:

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
$ sftp "esp@cbc"
esp@cbc's password:

Disconnected; connection lost (Connection closed.).


Warning: child process (/sys$system/tcpip$ssh_ssh2) exited with code 131.


%TCPIP-E-SSH_FC_ERROR, error in ssh file transfer operation

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

I have seen reports of this error happening during large file transfers but not during the initial login. From these reports I have gleaned that it is not a user quota problem but I have not seen any solutions.

Anyone have any ideas ?

[Allan Large]

Another fact I discovered. On August 20, 2010 an AUTOGEN was done with a SETPARAMS. The system had not been rebooted since that date until Nov 2 when a power failure caused a reboot.

The old params files have long been purged so I can't tell what all the system parameters were before. However, looking at the AGEN$PARAMS.REPORT file I see where NPAGEDYN was changed:

Old value was 6332416, New value is 11321344

Other than that, I see no significant changes. Could this be an issue ?

[labadie_1]

Check if you have some npagedyn allocation failure,

$ mc agen$feedback
$ sea sys$system:agen$feedback.dat NPAGEDYN_ALLOCFAIL

The first command just gives the prompt.

Can you post the output of the second command ?

[Allan Large]

$ searc sys$system:agen$feedback.dat npagedyn_allocfail
NPAGEDYN_ALLOCFAIL = 0
NPAGEDYN_ALLOCFAILPAGES = 0

[labadie_1]

So it does not seem to be relevant.

This problem is fixed in TCPIP 5.6 if memory serves me.

Check if you have the latest ECO for TCPIP 5.4, or upgrade to 5.6

[Allan Large]

I saw where the SSH_ALLOC_ERROR was fixed in 5.6 but in the release notes it said it was related to transferring large files. This is just the initial connection.

It is strange that SFTP had been working for at least 2 years (when EC0 7 was applied) until the system was rebooted after an AUTOGEN .

[Volker Halle]

Allan,

when AUTOGEN sets new parameters, it saves the old (previous) parameters in SYS$SYSTEM:ALPHAVMSSYS.OLD (see the 2nd line in SETPARAMS.DAT).

This gives you the chance to compare all system parameters from BEFORE the last AUTOGEN and the current parameters.

Volker.

[abrsvc]

CAn you post hte other changes that were made with the autogen pass? The change may not seem significant, but...

Thanks,
Dan

[Hoff]

sftp -vvv user@host for gonzo diagnostics.

check the ssh (sftp) server logs for errors.

verify the network is operational via ping and ssh.

verify minimum documented process quotas on client and on server.

verify that the target directories exist, that the login directories are OK, that the LOGIN.COM and SYLOGIN.COM procedures are not tossing errors, and that all associated devices are mounted and online, etc.

load current patches for VMS and for TCP/IP Services.

reverify the problem.

nb: HP support will ask for the current patches as part of standard call-handling procedures, so preemptively loading the patches helps ensure a faster trip through the HP support organization.

then call HP support.

[Allan Large]

abrsvc:

Parameter information follows:
------------------------------

MAXPROCESSCNT parameter information:
Feedback information.
Old value was 725, New value is 580
Maximum Observed Processes: 334

Information on OpenVMS executable image Processing:

Processing SYS$MANAGER:VMS$IMAGES_MASTER.DAT
Total global pagelets counted = 50194
Total global sections counted = 212
Total global pagefile counted = 329
Total resident code pages counted = 459
Total resident data pages counted = 0


GBLPAGES parameter information:
Feedback information.
Old value was 3246199, New value is 3484262
Maximum used GBLPAGES: 282112
Global buffer requirements: 196608
Pagelets reserved for memory resident sections: 0

GBLSECTIONS parameter information:
Feedback information.
Old value was 900, New value is 910
Current used GBLSECTIONS: 758
Memory resident sections: 0

LOCKIDTBL parameter information:
Feedback information.
Old value was 1792, New value is 13056
Current number of locks: 11155
Peak number of locks: 24576

RESHASHTBL parameter information:
Feedback information.
Old value was 2048, New value is 16384
Current number of resources: 4406

MSCP_BUFFER parameter information:
Feedback information.
Old value was 1024, New value is 1024
MSCP server I/O rate: 0 I/Os per 10 sec.
I/Os that waited for buffer space: 0
I/Os that fragmented into multiple transfers: 0

SCSCONNCNT parameter information:
Feedback information.
Old value was 5, New value is 5
Peak number of nodes: 1
Number of CDT allocation failures: 0

SCSRESPCNT parameter information:
Feedback information.
Old value was 300, New value is 300
RDT stall count: 0

SCSBUFFCNT parameter information:
Feedback information.
Old value was 50, New value is 50
CIBDT stall count: 0

NPAGEDYN parameter information:
Feedback information.
Old value was 6332416, New value is 11321344
Maximum observed non-paged pool size: 12984320 bytes.
Non-paged pool request rate: 2 requests per 10 sec.

LNMSHASHTBL parameter information:
Feedback information.
Old value was 512, New value is 1024
Current number of shareable logical names: 1989

ACP_DIRCACHE parameter information:
Feedback information.
Old value was 1446, New value is 1446
Hit percentage: 100%
Attempt rate: 204 attempts per 10 sec.

ACP_DINDXCACHE parameter information:
Feedback information.
Old value was 361, New value is 361
Hit percentage: 100%
Attempt rate: 152 attempts per 10 sec.

ACP_HDRCACHE parameter information:
Feedback information.
Old value was 1446, New value is 1446
Hit percentage: 93%
Attempt rate: 249 attempts per 10 sec.

ACP_MAPCACHE parameter information:
Feedback information.
Old value was 361, New value is 361
Hit percentage: 92%
Attempt rate: 0 attempts per 10 sec.

PAGEDYN parameter information:
Feedback information.
Old value was 6823936, New value is 7389184
Current paged pool usage: 2956336 bytes.
Paged pool request rate: 92 requests per 10 sec.

GH_EXEC_CODE parameter information:
Feedback information.
Old value was 1024, New value is 1536

GH_EXEC_DATA parameter information:
Feedback information.
Old value was 512, New value is 304

GH_RES_CODE parameter information:
Feedback information.
Old value was 1024, New value is 1024

GH_RES_DATA parameter information:
Feedback information.
Old value was 0, New value is 0

VAXCLUSTER parameter information:
- AUTOGEN parameter calculation has been overridden.
The calculated value was 1. The value 0
will be used in accordance with the following requirements:
VAXCLUSTER has been specified by a hard-coded value of 0.