Project

General

Profile

Actions
Copy link

Bug #632

closed

tcp.reassembly_memuse incorrectly reported

Added by Ludovico Cavedon about 12 years ago. Updated over 8 years ago.

Status:
Closed
Priority:
Normal
Assignee:
-
Target version:
-
Affected Versions:
Effort:
Difficulty:
Label:

Description

tcp.reassembly_memuse seems to be incorrectly multipled by the number of detector threads.

The static variable ra_memeuse holds the amount of memory used by the TcpSegmentPool, which is shared across all threads.
This global variable is used to set the value of the rtv->counter_tcp_reass_memuse counter (in StreamTcpReassembleMemuseCounter()), which is in the per-thread context.

However, when tcp.reassembly_memuse is printed on stats.log it is summed across all threads, therefore reporting the actual value of ra_memuse multipiled by the number of detector threads.

Verified on version suricata 1.3.4.

Actions
Copy link
 #1

Updated by Victor Julien almost 12 years ago

  • Status changed from New to Assigned
  • Assignee set to Anoop Saldanha
Actions
Copy link
 #2

Updated by Victor Julien about 11 years ago

  • Target version set to TBD
Actions
Copy link
 #3

Updated by Ken Steele about 10 years ago

Now (suricat 2.0.x) stats.log reports the value of ra_memuse for each thread. So it is no longer summed incorrectly, but it no implies that threads * ra_memuse amount of memory is being used, which is wrong.

Actions
Copy link
 #4

Updated by Victor Julien over 8 years ago

  • Status changed from Assigned to Closed
  • Assignee deleted (Anoop Saldanha)
  • Target version deleted (TBD)

This has been addressed in 3.0 I believe.

Actions
Copy link

Also available in: Atom PDF