We have a server which had a bad disk (out of 24). Trafficserver was
stopped and restarted before the disk was swapped (the disk has been marked
offline for a week). This is on version is 5.3.2 with some patches.
Stop, cleared the cache and restarted. Problem is gone.
Once it was restarted, I kept seeing this with "perf top" and CPU usage is
100% with load average of 1. Every request is a MISS.
Anybody seen this before?
Been going for more than 24 hours and server has been idle :
50.74% traffic_server [.] dir_clean_segment
30.65% traffic_server [.] dir_delete_entry
If I look a disk IO, there is nothing going on. Looks like stuck in a loop.
Enabling debug produced the following :
[Nov 1 16:06:31.127] Server {0x2b305e827700} DEBUG: (dir_clean) cleaning
0x2b372a76a838 tag 1A boffset 0 b 0x2b372a76a838 p (nil) l -1
[Nov 1 16:06:31.127] Server {0x2b305e827700} DEBUG: (dir_clean) cleaning
0x2b372a76a838 tag 1E boffset 0 b 0x2b372a76a838 p (nil) l -1
[Nov 1 16:06:31.127] Server {0x2b305e827700} DEBUG: (dir_clean) cleaning
0x2b372a76a838 tag 2E boffset 0 b 0x2b372a76a838 p (nil) l -1
[Nov 1 16:06:31.127] Server {0x2b305e827700} DEBUG: (dir_clean) cleaning
0x2b372a76a838 tag 32 boffset 0 b 0x2b372a76a838 p (nil) l -1
[Nov 1 16:06:31.127] Server {0x2b305e827700} DEBUG: (dir_clean) cleaning
0x2b372a76a838 tag 3A boffset 0 b 0x2b372a76a838 p (nil) l -1
[Nov 1 16:06:31.127] Server {0x2b305e827700} DEBUG: (dir_clean) cleaning
0x2b372a76a838 tag 3E boffset 0 b 0x2b372a76a838 p (nil) l -1
[Nov 1 16:06:31.127] Server {0x2b305e827700} DEBUG: (dir_clean) cleaning
0x2b372a76a838 tag 42 boffset 0 b 0x2b372a76a838 p (nil) l -1
[Nov 1 16:18:31.102] Server {0x2b305e827700} DEBUG: (dir_clean) cleaning
0x2b372a76a838 tag 171 boffset 0 b 0x2b372a76a838 p (nil) l -1
