The system was completely out of memory and had to start killing processes. Process "texis" being killed that way may cause database corruption depending on what it was doing when killed. You're probably running too many big walks at the same time.
The tech support info page will give you a rundown of memory usage.
On one of the 2 boxes there were 2 rewalks running (both replicating locally and remotely)
On the second there was NO rewalk running locally. Only the receiving profile was being updated.
Currently there is an updateindex script running which uses ~80% of physical memory.
Question? How many walks with what sizes is it safe to run at the same time?
That will require some math.
Consider: Total system RAM, "Parallelism: Servers", and "Max Process Size". Each parallel server can use up to "Max Process Size".
For each profile multiply it's "Parallelism: Servers" times that profile's "Max Process Size". That's the rough memory requirement for that profile.
Do not exceed total system RAM. It would be better to be a little under to leave room for caching etc.
Process sizes:
Small = 25MB
Medium = 50MB
Large = 100MB
Huge = 700MB
If you don't know how much memory you purchased in your appliance you can find out from the system log called "messages". Look for "kernel: Memory:". If you've booted recently it will be near the end of the log. If not it may have been rotated into messages.1, or messages.2 etc.
p.s.
Sometimes the "Remove common" feature can use more than the max process size amount of memory. It runs after the walk is complete but before the search index is created/updated.
