Holes on graphs after upgrade from 0.8.7e to 0.8.7i

[grigorov]

Hi, sorry for my english.

We are using cacti to collect data from 1530 devices (83732 ds, 50792 graphs). It was version 0.8.7e. We updated it to 0.8.7i and after that on every graph there are holes. What data i must provide for the next step.

Poller starts every 1 minute.

As example one attachment.

[Gatorz]

i have a plan to upgrade from 0.8.7e to 0.8.7.i also.
can we also go directly just install the latest one or have to do one by one ?
and do i need to upgrade the Spine also ?

[grigorov]

May be it will more clear if i post some configuratons.

First of all, when we used 0.8.7e MySQL engine was MyISAM, after upgragde i converted all tables to InnoDB. In update process there were error updating table 'data_template_rrd' table, so i changed 'cacti-0.8.7i-PIA-3.1/install/0_8_7g_to_0_8_7h.php'

before:
db_install_execute("0.8.7h", "ALTER TABLE `data_template_rrd` ADD UNIQUE INDEX `duplicate_dsname_contraint` (`local_data_id`, `data_source_name`, `data_template_id`)");

after:
db_install_execute("0.8.7h", "ALTER IGNORE TABLE `data_template_rrd` ADD UNIQUE INDEX `duplicate_dsname_contraint` (`local_data_id`, `data_source_name`, `data_template_id`)");

MySQL config:

# The MySQL server
[mysqld]
port = 3306
socket = /var/lib/mysql/mysql.sock
log_slow_queries = /var/lib/mysql/mysql-slow.log
max_connections = 50000
default-character-set = utf8
skip-locking

net_buffer_length = 8K
bulk_insert_buffer_size = 64M
key_buffer = 1G
max_allowed_packet = 16M
join_buffer_size = 128M
sort_buffer_size = 16M
read_buffer_size = 16M
read_rnd_buffer_size = 32M
myisam_sort_buffer_size = 128M
thread_stack = 256K
thread_cache_size = 256
thread_concurrency = 16
tmp_table_size = 512M
max_heap_table_size = 512M
table_cache = 128000
query_cache_limit = 16M
query_cache_size = 512M

innodb_data_home_dir = /var/lib/mysql/
innodb_data_file_path = ibdata1:10M:autoextend
innodb_log_group_home_dir = /var/lib/mysql/
innodb_log_arch_dir = /var/lib/mysql/
# You can set .._buffer_pool_size up to 50 - 80 %
# of RAM but beware of setting memory usage too high
innodb_buffer_pool_size = 1G
innodb_additional_mem_pool_size = 16M
# Set .._log_file_size to 25 % of buffer pool size
innodb_log_file_size = 256M
innodb_log_buffer_size = 8M
innodb_flush_log_at_trx_commit = 1
innodb_lock_wait_timeout = 50
innodb_file_per_table

All *.rrd files are stored in memory (tmpfs), MySQL DB files are on SSD disk.

[grigorov]

The problem is solved.

Table Engine was changed back to MyISAM in MySQL.