Boost updates and graph gaps

[ARC1450]

I have Boost installed (2.4) along with Spine for the poller.

Every time Boost runs an RRD update, I notice the NIC on my polling machine has less traffic coming in/going out, and there are gaps when the Boost update starts.

Just curious if there are some suggestions for tweaking that I could do. I noticed that either Gandalf or TheWitness said in another thread that Windows had an issue with waiting for one process to complete before starting another. I've also read some of the "poller output table not empty" threads and upped the connections to my MySQL DB along with upping some of the memory for query cache, tmp_table_size and key_buffer_size.

The specs of this machine are a Core Duo @ 3GHz, 3.75GB of RAM, and a 150GB HDD. I know slapping this on a machine with a better HDD would probably help things out, but I have to justify that, and my organization is a little OSS shy, so I'm trying to keep this under wraps for the most part.

[kevinmu0206]

The following is my "my.cnf" of MySQL, I used to have the same issue as you, but after I tweak MySQL configuration, my problem is solved, please adjust the following variable according to your hardware limitations.

[mysqld]
datadir=/var/lib/mysql
socket=/var/lib/mysql/mysql.sock
user=mysql
skip-innodb
skip-locking
skip-external-locking

# MySQL 4.x has query caching available.
# Enable it for vast improvement and it may be all you need to tweak.
query_cache_limit = 1M
query_cache_size = 128M
query_cache_type = 1

# max_connections=500
# Reduced to 200 as memory will not be enough for 500 connections.
# memory=key_buffer+(sort_buffer_size+read_buffer_size)*max_connections
max_connections = 300
key_buffer = 1024M

# myisam_sort_buffer_size used for ALTER, OPTIMIZE, REPAIR TABLE commands.
myisam_sort_buffer_size = 64M

join_buffer_size = 1M
read_buffer_size = 1M
sort_buffer_size = 16M

#tmp_table_size=32M by default
tmp_table_size = 64M
table_cache = 1024

# 1Mb of read_rnd_buffer_size for 1GB RAM — see notes below.
# read_rnd_buffer_size=256K by default.
read_rnd_buffer_size = 2M

# thread_concurrency = 2 * (no. of CPU)
thread_concurrency = 8
thread_cache = 200
thread_cache_size = 256

# Reduced wait_timeout to prevent idle clients holding connections.
wait_timeout = 180
connect_timeout = 10

max_allowed_packet = 16M
max_connect_errors = 10

# Default to using old password format for compatibility with mysql 3.x
# clients (those using the mysqlclient10 compatibility package).
old_passwords=1

# To allow mysqld to connect to a MySQL Cluster management daemon, uncomment
# these lines and adjust the connectstring as needed.
#ndbcluster
#ndb-connectstring="nodeid=4;host=localhost:1186"

[mysqld_safe]
log-error=/var/log/mysqld.log
pid-file=/var/run/mysqld/mysqld.pid

[mysqldump]
quick
max_allowed_packet = 16M

[myisamchk]
key_buffer = 64M
sort_buffer = 64M
read_buffer = 16M
write_buffer = 16M

[TheWitness]

You definitely need to be more specific. A few screen shot's would help. Boost Status page would help. Also, the section of the Cacti Log that is applicable to the Gaps. I expect the color in the Log to be 100% green.

TheWitness

[ARC1450]

Okay, so I ended up switching over to the ENGINE=MEMORY version of Boost. . .works better.

The only problem is that 3.0 Boost chews up a ton of RAM. I'm still going through that to figure out what's happening.

I just used the settings from 2.4 Boost that worked just fine, but on 3.01 they don't.

Eh well, I shall have to tweak the DB I think. I noticed Witness offered up an alter table to ease up on RAM usage, so I may have to toss that out there to fix the issue.