Tranquil IT Forum

Hello,

Starting today, when I open the WAPT console (Enterprise version), I get the following error message:

Unable to get hosts list: Error on server:
InternalError('ERROR: missing chunk number 0 for toast value 5546857 in pg_toast_16387\n',)

Clicking OK doesn't display the list of machines.
However, if I click on one of the Active Directory subfolders, the machines appear...
So the problem only occurs when I select "(All)"...

Do you have any ideas?

The TOAST tables in Postgres store BLOBs, which are large strings of characters like JSON inventory reports. The standard Postgres database tables (i.e., the non-TOAST tables) reference the objects stored in the TOAST tables. And here, the message clearly implies an inconsistent reference... which is more of a Postgres issue than a WAPT issue.

Was there a sudden outage that could have corrupted the database? Is this on a Windows or Linux server? What version of WAPT? What version of Postgres?

It's possible to clean up the corruption in question (see https://gist.github.com/supix/80f9a6111 ... b9dedf187a) , but the simplest solution is probably to restore a backup from the previous day. The workstations will automatically restore their inventory once the server is back online.

That said, it's really not normal for this to happen; the entire storage layer, starting with the filesystem, should be checked.

Denis

You can check this page to see if there's a condition that applies to your installation : https://wiki.postgresql.org/wiki/Corruption

Hello Denis,

Thank you for your reply.

I finally opted to restore the virtual machine on the Windows server hosting the WAPT server.

To answer your questions:
- Was there a sudden outage that could have corrupted the database?
No, not to my knowledge.
- Is it on a Windows server or a Linux server?
Windows server (Windows Server 2019)
- Which version of WAPT?
1.8.1.6756
- Which version of Postgres?
I have no idea where I found that.

Thank you.
Christophe.

Okay, but it's really not normal that the Postgres database has become corrupted... There must be something strange going on, probably at the OS and/or virtualization level.

Yes, especially since this is the second time this kind of problem has occurred...

We have no issues with our other 14 servers (including database servers).

Before the restore, what's most surprising is that the machine list displayed correctly if I went into each of the subdirectories in the left-hand list... The error only appeared if I selected the "(All)" view... Yet I have the impression that it's simply a different way of displaying the machines...

Nevertheless, the error message refers to a Postgres corruption, which WAPT can do little about... Otherwise, you would need to consult a Postgres expert.