VSI OpenVMS Forum

We have a repository hosted on our local LAN that is being transitioned to github.com. The local LAN repository could be accessed via ssh. Now that it's on github, we need to access it through a proxy since github.com isn't directly reachable.

I've changed git config file on OpenVMS to point to the new origin.

[remote "origin"]
url = https://github.com/xxxxxx

I've add the proxy entry also:

[http]
proxy = http://nnn.nnn.nnn.nnn:pppp

However, whenever I try to pull, push, or otherwise access the repository I get the error:

Bad news:
SSL error: syscall failure: bad file number [2]

which seems to indicate it is ignoring the proxy and trying to contact https://github.com directly.

Any ideas on how to solve this?

The alternative is to clone the repository on my desktop, and sftp changed files to/from OpenVMS to my desktop to work with github, since the proxy works there.

I've now installed VGIT V1.4-1B in the hopes that it fixes the proxy problem.

The first time a tried vgit status, it racked up thousands of I/O's and then actually crashed my x64 system with a bugcheck:

CPU bugcheck codes:
CPU 00D -- database address 893ADB00 -- UNXSIGNAL, Unexpected signal name in ACP

CPU 00D reason for Bugcheck: UNXSIGNAL, Unexpected signal name in ACP
Process currently executing on this CPU: CSEWELL
Current image file: DAB911$DKA0:[SYS0.SYSCOMMON.][SYSEXE]VGIT2.EXE;1

That's not good. So from scratch I set up the proxy with:

vgit2 config http.proxy http://nnn.nnn.nnn.nnn:8080

Then I tried cloning a repository from github

vgit2 clone https://github.com/company/repository

It asked for username and password and seemed to accept them; it created the directory and subdirectories, etc. and then spit out the error:

Bad news:
SSL error: unknown error [16]

which apparently means the SSL certificate is invalid. Any ideas on how to solve this?

csewell wrote: ↑

Thu Jan 18, 2024 2:50 pm

The first time a tried vgit status, it racked up thousands of I/O's and then actually crashed my x64 system with a bugcheck:

CPU bugcheck codes:
CPU 00D -- database address 893ADB00 -- UNXSIGNAL, Unexpected signal name in ACP

CPU 00D reason for Bugcheck: UNXSIGNAL, Unexpected signal name in ACP

Could you please provide the crash footprint file (from CLUE$OUTPUT:CLUE$node_ddmmyy_hhmm.LIS) of this crash as an attached file ?

If you can, please log a call for this crash.

Volker.

Yes, please enter a support ticket for this with enough info for us to reproduce the failure.

What version of the OS are you running?

Do you have GIT$SSL_CERTS logical? It should points to SYS$MANAGER:CA-BUNDLE.CRT

Hi, have you been able to make up bug report for us with some info about that crash?

Support ticket created. Not sure I feel comfortable adding a CLUE file to an open forum, but I did add them to the support ticket. The server is not a hobby server.

Added in 3 minutes 47 seconds:
I just did a "vgit status" and crashed my server again. I won't be doing that again until this problem is addressed.

Added in 6 hours 37 minutes 18 seconds:
Apparently the usual support ticket channel is only for reporting OpenVMS X86 problems. I suppose I'll just email VSI support instead.

> Support ticket created. Not sure I feel comfortable adding a CLUE file to an open forum, but I did
> add them to the support ticket. The server is not a hobby server.

Support tickets are not posted to an open forum; only VSI employees see them.


-- Rob