Mailing List ecs-isp@2rosenthals.com Archived Message #296

Fra: "Lewis G Rosenthal" <ecs-isp@2rosenthals.com> Full Headers
Undecoded message
Emne: Re: [eCS-ISP] links in the ticket
Dato: Mon, 29 Mar 2021 15:47:09 -0500
Til: eCS ISP Mailing List <ecs-isp@2rosenthals.com>

Hi...

On 03/28/21 10:15 pm, Steven Levine wrote:
In <list-1764398@2rosenthals.com>, on 03/28/21
    at 11:05 PM, "Lewis G Rosenthal" <ecs-isp@2rosenthals.com> said:

Hi again,

Recall that the bootstrap check is only to ensure that a YUM/RPM
subsystem  exists.
This might the goal, but IMO, this is not what it actually does.  It
checks for the presence of two executables which may or may not run and
which may or may not be in the \usr tree.


I had to refresh my recollection of the code. Indeed, you are correct. To this point, this is all that's really been required, but I can see how it may not be sufficient for all cases.

I'm not advocating for any kind of repair option.  The vast majority of
ANPM installs run just fine and the current repair facilities seem to get
the job done.

What I might find useful, in rare cases, is a /BOO command line option to
force a bootstrap install when the user has put the system into a state
that manages to fool ANPM bootstrap check.


And what would that do in practice? Assuming the system has:

\etc
\usr
\var\cache
\var\lib
\var\log\yum.log
\var\run
\var\tmp

would we download the bootstrap, unlock all files in those directories, delete them, and then unzip the bootstrap before prompting for a reboot? Recall that this is essentially what my script does when I am updating the bootstrap content (except that it renames all of those, which takes up considerably more space after the latest bootstrap is extracted for updating).

Because re-bootstrapping is potentially a destructive operation, we need to be clear as to what we're doing and why. Perhaps the reason that yum might not be working is that yum.conf is broken; re-bootstrapping becomes great overkill.

I'm just playing devil's advocate.

In Massimo's case he corrected the platform selection and probably has
installed enough packages correctly by this time so that there's no longer
any real need to force a bootstrap install.

That would be my guess. Not many current packages are in the i386 repo.

Perhaps the wiki needs a
new page concerning the  nuclear option.
The content is pretty much already there although some consolidation might
help.  What's missing IIRC is some discussion of when forcing a bootstrap
is the best option and what the user should try before forcing the
bootstrap.

Roger that. That's an item for the troubleshooting page.

I've really tried to avoid the
subject, but I guess there  are times when the elephant gun approach is
simply the shortest distance  between the two points (non-functional ->
functional).
I understand, but is it a better solution to write it once or have to
write it multiple times in support tickets or one some mailing list.


Point taken.

I just don't want to encourage the nuclear option as a *first* resort rather than a *last* resort, as people will invariably lose configuration data or package combinations due to lack of thorough/current backups...and then blame the messenger.

In truth the reason you have been able to avoid documenting it in th wiki
is that it is rarely required.  For some reason it seems that user's need
to reinstall the OS more often than they need to recover just the
ANPM/rpm/yum controlled content.


Yes.

--
Lewis
-------------------------------------------------------------
Lewis G Rosenthal, CNA, CLP, CLE, CWTS, EA
Rosenthal & Rosenthal, LLC                www.2rosenthals.com
visit my IT blog                www.2rosenthals.net/wordpress
-------------------------------------------------------------


Abboner: Feed, Digest, Index.
Stopp abbonement
E-post til ListMaster