No packages installed/available, and bad pkg mgr links? HELP!

MrPete

I'm quite confused.

I'm running 2.6.0-RELEASE (amd64) built on Mon Jan 31 19:57:53 UTC 2022

The system has been mostly stable, but I needed to do some work on ACME certs so started poking around... only to find some rather confusing symptoms:

The package manager says I have no packages installed
And can't retrieve any package info

I began going through the Broken Pkg Database troubleshooter. Before running the mkdir, I took a look at the folders involved, and found this:

#ls -l /var/db/pkg
lrwxr-xr-x  1 root  wheel  21 May 21 09:41 /var/db/pkg -> ../../root/var/db/pkg
#ls -l /root/var/db/pkg
lrwxr-xr-x  1 root  wheel  21 Mar 28 14:58 /root/var/db/pkg -> ../../root/var/db/pkg

That sure looks like a circular link loop LOL

QUESTIONS:
Does this look normal?
Where should the pkg database be?
Any idea how to fix this, and is there something I can do to help diagnose before we fix? I'd hate for this to happen to others.

bingo600

@mrpete
That looks odd (Bad)

I don't have a 2.6.0 , but this is from a 22.01 (plus)

[22.01-RELEASE][admin@fw01]/root: ls -l /var/db/pkg
total 3611
-rw-r--r--  1 root  wheel  6520832 Apr 25 14:48 local.sqlite
-rw-r--r--  1 root  wheel      158 Feb  7 19:20 pfSense-core.meta
-rw-r--r--  1 root  wheel      158 May 12 16:31 pfSense.meta
-rw-r--r--  1 root  wheel   221184 Feb  7 19:20 repo-pfSense-core.sqlite
-rw-r--r--  1 root  wheel  1253376 May 12 16:31 repo-pfSense.sqlite
[22.01-RELEASE][admin@fw01]/root: ls -l /root/
total 24
-rw-r--r--  2 root  wheel  1023 Feb  7 17:56 .cshrc
-rw-r--r--  1 root  wheel     0 Apr 25 14:47 .hushlogin
-rw-r--r--  1 root  wheel    80 Feb  7 17:56 .k5login
-rw-------  1 root  wheel    38 May 16 20:23 .lesshst
-rw-r--r--  1 root  wheel   328 Feb  7 17:56 .login
-rw-r--r--  2 root  wheel  1140 Apr 25 14:47 .profile
-rw-r--r--  1 root  wheel  1943 Apr 25 14:47 .shrc
-rw-r--r--  1 root  wheel  3090 Apr 25 14:47 .tcshrc

Btw:
How do you get that # prompt ?
Do you have sudo installed ?

/Bingo

MrPete

@bingo600 (The # I just inserted in place of the real prompt, rather than delete machine name/etc details ;) )

THANKS!

Sounds like somehow my pkg database really did get wiped.

Gonna look around for evidence if possible...

MrPete

@jimp Perhaps this is actually in your wheelhouse:

I just noticed a redmine, #13182 saying that having RAM disk (for /tmp and /var) conflicts with ZFS.

I upgraded to 2.6 via reinstall/restore to get ZFS. Didn't know about this RAM disk bug. (Maybe wouldn't do ZFS if I had known: running on SSD, it's not a great idea to have high volume SSD traffic for tmp files and such)

Could having RAM disk enabled be the source of corruption on my pkg database?

bingo600

@mrpete
I have been running ZFS since 2.4.5 on SSD, wo. any issues at all.
I don't think the new SSD disks are that sensitive anymore, new wear level algo's , and better cells.

Well i did avoid the Samsung QVO's, and went for EVO's.

/Bingo

MrPete

@bingo600
Wearout still does happen after a bunch of TBW.

(The other concern nobody talks about: bitrot due to read without overwrite. Life is about a year for consumer SSD's, worse for enterprise. Easy workaround: use a utility to overwrite every sector in use about 2-3 times a year.)

MrPete

@bingo600 GOOD NEWS. Solved it, and it IS due to a bug of RAM disk + ZFS.

See here.