Discussion related to cado-nfs

[Ed Hall <ed_ka2fwj@yahoo.com>]

----- Forwarded Message -----

From: Ed Hall <ed_ka2fwj@yahoo.com>

To: Emmanuel Thomé <emmanuel.thome@inria.fr>

Sent: Sunday, August 23, 2020, 9:24:31 PM EDT

Subject: Re: [Cado-nfs-discuss] Clients not getting WUs

Thanks for your help, Emmanuel,

I've been running CADO-NFS across >30 machines for a few years now and don't remember the exact failure, but last year my Core2 machines wouldn't compile with the then current commit.

Tonight I compiled the latest commit without incident on one of my Core2 machines and it is working fine with the server which is using yesterday's commit.  Of course, the rest of the "farm" will need to be updated, but all looks fine ATM.

Thanks much for all.

Sincerely,

Ed

On Sunday, August 23, 2020, 9:12:14 AM EDT, Emmanuel Thomé <emmanuel.thome@inria.fr> wrote:

On Sun, Aug 23, 2020 at 12:23:38PM +0000, Ed Hall wrote:
>  Thank you, Emmanuel.  I have many machines, some of which are Core2 that didn't seem to work with commits later than some time in 2018.  Is Core2 supported in these later commits?  Maybe I was really having a different issue with them?

Of course "core2" is supported (whatever core2 means, since commercial
CPU names have been totally meaningless for well over a decade).

Could it be that with your setup you accidentally fell into the trap of
attempting to run server-shipped binaries on incompatible clients ?

If yes, please read:

    https://gitlab.inria.fr/cado-nfs/cado-nfs/#run-a-larger-factorization-on-several-machines
    https://lists.gforge.inria.fr/pipermail/cado-nfs-discuss/2020-March/001168.html

Cheers,

E.

> I will work on upgrading all the others and see how my "farm" does.(smile)
> Thank you everyone for all the work!
> Sincerely,Edwin Hall
>
>    On Sunday, August 23, 2020, 6:02:53 AM EDT, Emmanuel Thomé <emmanuel.thome@inria.fr> wrote: 
> 
>  Yes please. I mean, it's not a stringent requirement in general, but for
> the matter at hand, you're indeed crossing an incompatibility boundary
> (or so I guess).
>
> STDOUT was added in WUs as a consequence of commit a4fc0715dc (march
> 2020), which provides a fix to https://gitlab.inria.fr/cado-nfs/cado-nfs/-/issues/21718
>
> So any pre-march2020 client will choke on WUs sent out from a
> post-march2020 server. Sorry for that.
>
> Cheers,
>
> E.
>
> On Sat, Aug 22, 2020 at 09:05:28PM +0000, Ed Hall wrote:
> > Hi Team,
> > I'm trying to use the latest commit that was made today and the server is sending out WUs and then waiting for timeout:
> > Info:HTTP server: 192.168.0.77 Sending workunit alicado_polyselect1_36000-40000 to client alicado99.77
> > But, the clients are showing Errors:
> > 2020-08-22 16:52:21,906 - INFO:root:Opened URL http://math99.local:13531/cgi-bin/getwu?clientid=alicado99.77 after 290.0 seconds wait
> > 2020-08-22 16:52:21,906 - ERROR:root:Invalid workunit file: Error: key STDOUT not recognized
> > 2020-08-22 16:52:21,907 - INFO:root:Removing workunit file download/WU.alicado99.77
> > 2020-08-22 16:52:21,907 - INFO:root:Downloading http://math99.local:13531/cgi-bin/getwu?clientid=alicado99.77 to download/WU.alicado99.77 (cafile = None)
> >
> > Do I need to run the same commit on the clients for this revision to work?
> > I have saved a copy of the working directory, but I'm not sure where to upload/attach which files.
> >
> > Sincerely,Edwin Hall
>
> > _______________________________________________
> > Cado-nfs-discuss mailing list
> > Cado-nfs-discuss@lists.gforge.inria.fr
> > https://lists.gforge.inria.fr/mailman/listinfo/cado-nfs-discuss
>
>