Closed
Description
Percona has additions in innodb output:
-------------- ROW OPERATIONS -------------- 0 queries inside InnoDB, 0 queries in queue 1 read views open inside InnoDB 1 transactions active inside InnoDB 1 out of 1000 descriptors used ---OLDEST VIEW--- Normal read view Read view low limit trx n:o EA05374F2 Read view up limit trx id EA05374F1 Read view low limit trx id EA05374F2 Read view individually stored trx ids: Read view trx id EA05374F1 ----------------- Main thread process no. 28092, id 140035932575488, state: flushing log Number of rows inserted 47675820507, updated 5833120651, deleted 10936909882, read 501029518850 0.00 inserts/s, 43442.28 updates/s, 0.00 deletes/s, 76878.06 reads/s ------------ TRANSACTIONS ------------
In this case the line '-----------------' after OLDEST VIEW is treated as beginning of chunk name and TRANSACTION as a chunk text.
Solution:
# diff -u innodb_preprocessor.py.orig innodb_preprocessor.py
--- innodb_preprocessor.py.orig 2015-10-30 05:39:18.073914726 -0700
+++ innodb_preprocessor.py 2015-10-30 07:37:29.439914287 -0700
@@ -39,19 +39,23 @@
chunks = {'junk': []}
current_chunk = 'junk'
next_chunk = False
+ oldest_view = False
self.clear_variables()
for row in rows:
innoblob = row[2].replace(',', '').replace(';', '').replace('/s', '').split('\n')
for line in innoblob:
# All chunks start with more than three dashes. Only the individual innodb bufferpools have three dashes
+ if line.startswith('---OLDEST VIEW---'):
+ oldest_view = True
if line.startswith('----'):
# First time we see more than four dashes have to record the new chunk
- if next_chunk == False:
+ if next_chunk == False and oldest_view == False:
next_chunk = True
else:
# Second time we see them we just have recorded the chunk
next_chunk = False
+ oldest_view = False
elif next_chunk == True:
# Record the chunkname and initialize the array
current_chunk = line
Metadata
Metadata
Assignees
Labels
No labels