​

1 Error log location —- The read service provider is offline

Read service taken offline:

​

2 Emergency Solution:

Save a machine’s gC. log log file, thread other log files, etc. And then restart the service

3 Check the offline cause:

  • Machine condition Physical memory 8G started two Java processes

  • ​

  • 1. Eliminate VM memory leaks through GC logs; The figure below

  • ​

  • 2 egrep -i “killed process” /var/log/messages -C100 OR Dmesg – T | grep Java – C100 confirm whether Linux OS killer out The second chapter figure above, is been OSkill indeed

  • ​

    3 Suspicious process DNF pull metadata update (only suspicious, two machines are running DNF recently, OS OOM occurred, resulting in only one provider left, finally failed to handle)

    ​

  • 4 Calculate the actual memory usage before the OOM

  • 2240 + 498 + 225 + 96 + 487 + 215 + 36 + 38 + 224 + 231 + 1808 + 137 + 698 + 429 + 397 + 632 + 3703 + 321 + 235 + 511 + 228 + 51 + 28 + 34 + 90 + 2305 + 328 + 715 + 410 + 2691 + 38 0+8699+563135+1249532+2112+82+594+86+86+25797+20+24+35+12451+5803+16+22=1888915 x 4kb=7.2 gb Java two processes + all other processes RSS has a total of 7.2g and should have 700M of memory, so you need to ask Google here, basically nothing can be explained clearly.

    ​

4 Solutions

  • Procedure 1 Run the following command to disable the DNF timer.
  • systemctl stop dnf-makecache.timer
  • systemctl disable dnf-makecache.timer
  • 2 Upgrade the memory
  • 3 (Alarm of existing business)

5 Summary & reference

  • ** Root cause: ** When Linux is running out of memory, the system will treat the process that occupies too much memory as a rogue process, and then send a signal to kill the rogue process, resulting in the Java application service cannot be used
  • DNF related command www.howtoforge.com/tutorial/ce…

​