Cacti 0.8.4 Out

[raX]

After weeks of testing, and over two months of preperation, 0.8.4 is out! This version is just the beginning of cool things to come, which includes the new template XML import/export support. Check it out .

Below are the complete changes:

Code: Select all

-bug#87: Deleting top level branches without children on the tree caused ordering
problems.
-bug#88: Possibly popen/pclose concurrency problem with cactid.
-bug#89: Missing "check all" checkbox on polling hosts page when an empty column
was drawn.
-bug#91: Fix undefined index errors on data source checkbox popups.
-bug#92: Fix undefined index errors on graph management checkbox popups.
-bug#96: Problems creating graphs from data queries where there was only row row total
in the query result.
-bug#97: Typo in cmd.php
-bug#98: Creating a data source or graph with no host or template, would result in an
extra empty item.
-bug#99: Deleting a graph or data template that was still attached to one or more
items would cause the graph/data source to appear to still have the deleted
template still attached.
-bug: Removed size limits on Linux memory data template.
-bug: Undefined variable errors when creating new graphs/data sources
without a template.
-bug: multiple problems that caused the Windows disk space and CPU graphs
to stop working.
-bug: Broken images for bandwidth summation graphs that were less than a day old.
-bug: Graph order changes on the tree would affect other graph trees.
-bug: Problem with the "Host MIB - Hard Drive Space" data template that broke
Windows disk spaces graphs.
-bug: Fixed the LDAP authentication code, which was partially broken in 0.8.3.
-bug: Fixed a cmd.php parsing problem when input and output fields are sharing the
same name.
-bug: Added basename() to all PHP_SELF references to make sure all hrefs stay
absolute.
-bug: The RRA field is now honored for the dual pane tree view.
-bug: The |date_time| graph variable now displays the current day of the month.
-feature: The "Total All Items" CDEF is now able to total NaN's without becoming
'U'.
-feature: Data query results are now grayed out if they have already been created.
-feature: The "cacti Web Root" and "Web Server Document Root" settings are completely
auto-detected now.
-feature: Add SNMP port and SNMP timeout fields to each host.
-feature: Removed the "Management IP" field in polling hosts in favor of "Hostname".
-feature: Re-organized the menus expand/contract for less used items to help save
vertical space.
-feature: Added complete template to XML import and export support.
-feature: Moved the graph creation features of "Polling Hosts" to a new menu item,
"New Graphs".
-feature: Added verbose debugging support for data queries.
-doc: Added a FAQ.
-cactid: Removed lots of potential buffer overflows.
-cactid: Several libz/openssl autoconf fixes that improve FreeBSD 5.x support.
-cactid: Fixed some potential segfaults on FreeBSD when reading information from
the targets table.
-cactid: The snmp_get() function now respects the SNMP version chosen under
"Polling Hosts", so 64-bit counters should work.

-Ian

[Guest]

Thank you, thank you, thank you!



*installing*

[egarnel]

Smoothest upgrade yet! took all of 5 minutes with no glitches.

Excellent job Rax!

[TFC]

Is upgrade procedure same?

[Yendor]

Wow you widened the ip field so now i can prefix it with tcp: to get it to use tcp instead of udp, cool

I still have to modify lib/snmp.php to use -A instead of -X to snmpget otherwise i get the error

Error generating a key (Ku) from the supplied privacy pass phrase.

but that isn't a big deal since I havn't bothered working out why I am getting that error.

The only other problem I can see is that if you run cacti in a frame in say your intranet (which we do) then for some reason the links to the tree items seem to take over the whole browser rather then staying in the same frame.

I have a question though, I noticed when looking at the man page that (at least on linux) net-snmp's snmpget lets you pass multiple oids to it to get at once. I would imagine that getting all the values at once would be faster then getting them one at a time, if for nothing else then simple for the extra overhead in invoking snmpget each time. Has this ever been tried before (ie getting all the data from a host in one hit) or is there some issue I don't know of that would make this a bad idea ?

Thanks for the hard work Rax