Test System now getting Segmentation Faults

[aegersz]

No worries But again, a PHP fatal error should not crash Apache...

well, it crashed Chrome but Firefox managed to limp along ... sort of.

it's my fault for incomplete rsyncing - a genuine mistake because i forgot maybe ?

but at least i learned something !

[LiroyvH]

I'd still keep looking for segfaults. Those aren't supposed to occur indeed and yes: it may point to a hardware issue or config problems.

[aegersz]

they came back as soon as i imported a fresh database.

maybe it's too big ? the number on line users is wrong and says 200+ ... another mystery.

[青山 素子]

As I've said on similar topics before, segmentation faults point to bugs in the server software or possibly the hardware. Nothing run at the high level in PHP or on MySQL should cause a segfault in normal usage. If it does, it's a problem with PHP/MySQL or similar.

[aegersz]

i think it's because i am moving the large database to a small laptop.

ok, thanks, i may well have to just live with it.

[青山 素子]

i think it's because i am moving the large database to a small laptop.

No, it isn't. That should never cause a segfault. There are a few things that can cause such an error including buffer or stack overflows, dereferencing null pointers, or accessing non-existent memory. These types of errors happen within the compiled program. They won't (normally) be caused by PHP code or SQL files.

For most large and popular programs, a segfault should be exceedingly rare. The typical causes in those cases stem from hardware issues or broken applications.

Simply importing a large database should never cause this type of issue. At most, it would just be very slow. I speak from experience, having migrated multi-gigabyte databases with several-million-row tables.

ok, thanks, i may well have to just live with it.

Please don't. Having that happen points to a larger issue and you don't want weird failures later on. You likely either have a bad build of MariaDB/MySQL you're using or some kind of hardware problem.

[aegersz]

yeah, it must be due to my cavalier relationship with my mirror/laptop - it's really old too, with not much memory.

i had been trying to install a new version of php and failed ... it might be related.

just as long as my live system is fine, i'm ok with that.

i may get segmentation faults that kills Chrome but Firefox sort of handles it better.

thank you for your help ... i may go down the laptop memory path next time ... it was working well too until the database grew - i know that you say it's not connected but whenever i restore the database on the laptop, the active user count keep accumulating into the 100's ! i can't figure out why and still think that it's the root of all my problems, somehow.

[aegersz]

Kindred explained this - it's in the LAST 15 minutes ... i see now.

thank you so much.

[aegersz]

I can avoid the problem by not mirroring the Sources directory and provided I manage the test system, post complete rebuild, all by itself, it's ok.

I just feel that there is a mod that's giving me this grief but it's only a guess.

by not being able to sync the Sources, my test system looks very basic but at least I can import databases.

something is lurking there that causes this yet my live system is going well.

[aegersz]

i rebuilt my mirror intranet system on a more modern laptop that uses SSD ... and rsync'ed all live files and the db.

it all now works like a dream.

i put it down to old hardware - this is now resolved.