D&C GLug - Home Page

[ Date Index ] [ Thread Index ] [ <= Previous by date / thread ] [ Next by date / thread => ]

Re: [LUG] Control Systems

 

On 08/04/2020 20:29, comrade meowski wrote:
On 08/04/2020 17:47, Simon Avery wrote:
    On 07/04/2020 15:33, Simon Avery wrote:

    Ah, I'm guessing it interpreted /etc/fstab and tried to mount the
    NFS shares before the local filesystem was ready.

    It's a nice trick that, but not exactly endearing

    No, it ignored fstab completely as far as I know. Of course you may
    be right, how would I fix that? Everything that worked perfectly
    well before systemd shut up shop. In the end I had to write a .mount
    file for every NFS share, then .automount files for each of them,
    and then put shortcuts to the shares on the desktop. Then after all
    that rigmarole I had the functionality back that I had before
    systemd arrived.

Apologies, I got that slightly wrong. It was NFS shared trying to be mounted /after/ the local file system, but /before/ network was ready.

systemd will convert all entries in /etc/fstab into .mount files automatically on boot - search for these with: systemctl | grep “.mount”

I think (not at work so can't check my notes) - I added something like: After = network-online.target

However - this was over a year ago. I don't think I changed our build scripts for this, and I've since build dozens of machines without doing this and even though we use NFS a lot, it's not something I've had a problem with for some time (hence my vagueness) - so I guess systemd has gotten a bit smarter about that particular thing.

Haha, I remember that episode very differently and I'm sure that grepping the DCGLUG mailing list history would support me as well. Predictably it was me who talked him through it back then...
It was indeed!
The problems were decidedly layer-8, "user" (sorry Julian).
No problem :) At work we had a resolution code: Registered -> Technical -> User. Otherwise known as a One Dee Ten Tee [1D10T] error or PEBCAK - Problem Exists Between Chair And Keyboard :) You're right anyway, I do have a tendency to try what I think is right, make a pig's ear of it, and it's one of you helpful folk who digs me out of the self inflicted manure (as you know all too well!).
Someone was determined to ignore my advice on doing things the "proper" systemd way even though it was working at his end very quickly - as you say setting up systemd to do NFS mounts is hardly difficult ¯\_(ツ)_/¯

Unfortunately a certain someone didn't "like" the way things were setup and insisted that the NFS drives were mounted automatically - forcibly might be the better term - during boot despite me repeatedly pointing out that they were already set to lazy mount transparently whenever he accessed them for any reason which is the same but better. It went backwards and forwards for ages I believe and eventually something yucky was bodged up and we all moved on.
In my defence what I wanted was just for the system to behave as it always had - anyone remember Garth from Wayne's World 'We fear change'? *lol* - e.g. NFS shares mounted during boot and easily accessible via handy desktop icons. I ended up with the -probably overly complicated - solution I mentioned above.

In fairness though, it's nobody's job to tell other people how they "must" run their own systems
No but there is a difference between 'I've always done it like this' ignoring better ways, and telling people they must /just/ because it's the new way, and I do accept that :)
and this is Linux after all - in a way he got the best possible solution which was a dodgy ghetto "it sucks but it works and I did it my damn self" type fix and ultimately I can't argue with that. I kind of like it when people ignore my advice, fix things themselves in weird ways and end up happy with it. Fair enough!

Looking sideways at my pile of computers and networking gear it's basically *all* like that anyway. "Let he who is without sin..."

Have very good reactions.. mine have been repeatedly honed :)

Kind regards,

Julian

--
“The great tragedy of Science — the slaying of a beautiful hypothesis by an ugly 
fact.”

― Thomas Henry Huxley


--
The Mailing List for the Devon & Cornwall LUG
https://mailman.dcglug.org.uk/listinfo/list
FAQ: http://www.dcglug.org.uk/listfaq