XDR ran out of queue space

xdr

#1

Hi guys,

We’re using Aerospike ver. 3.8.X, and saw the following line in the log

WARNING (xdr): (xdr.c:4999) Ran out of queue space… XDR cannot keep up with write … some records may be lost!!!

May I know what’s this queue, is it the ringbuffer digest log queue? And how to fix this warning?

Thanks!


#2

It is an internal queue that cache the data to be written to the digest log. So your digest log is probably full. In that case, you have two options. Either remove the digest log or use this command:

asinfo -v ‘xdr-command:skip-outstanding’


#3

Also check your disk activity using iostat and see how busy it is. May be its not able to take that much amount of writes.


#4

To clarify, when digest log is full, it will overwrite old data so I was not able to reproduce the “Ran out of queue space” issue. However, I was able to reproduce the issue when using a small filesystem to hold the digest log which is explained here with solution:


#5

Got it, thank you!


#6

Here is an article covering this (adding here for future reference): Ran out of queue space... XDR cannot keep up with write