lmxdbstat - reports status of LMX port and buffer traffic for databases

lmxdbstat [-v] [-?|-h] [-p <pid>] [-d <db1>[,<db2>...]] [ interval [count]]

lmxdbstat -z <0/1>

The lmxdbstat utility reports the status of the LMX protocol and shows statistical counters of port and buffer traffic for various databases. It is a wrapper script around lmxstat(1M) that has all the options supported by lmxstat with additional options for reporting LMX statistics for databases. This utility can be very helpful in getting the LMX statistics for a particular database thereby identifying the root cause for various database issues at the communication layer. Both port and buffer data transfers are done over LLT protocol. LLT currently supports interconnects like ethernet/GigE. lmxdbstat displays the output for both received and transmitted data in packets and Kbytes. With no options the output displays the received and transmitted data in packets and Kbytes per second. In the default mode lmxdbstat displays the output in 2 rows. The first row indicates number of packets received and transmitted for port and buffer traffic in the the last one second and also the total of both received and transmitted packets. The second row indicates number of Kbytes received and transmitted for port and buffer traffic in the last one second and also the total of both received and transmitted Kbytes.

-v	In the verbose mode the output is displayed as the average over the last 1, 10 and 30 second intervals. The first row indicates the packets of port and buffer traffic received and transmitted in the last one second and also the total of both of them. The second row shows the average number of packets of port and buffer traffic received and transmitted in the last 10 second interval and the third row shows the same information for the last 30 second interval. The next three rows show this data in Kbytes.
-?|-h	To show help or usage page for lmxdbstat
-p <pid>	To get statistics for a process with PID=<pid>
-d <db1>[,<db2>...]
	To get statistics for a database(s) <db1>[,<db2>...]. By specifying more than one database, comparison between them can be done easily.
-z <0|1>	To disable/enable LMX statistics collection. In disable mode, -p and -d options are not usable.

The following operands are supported:

count Specifies the number of times that the statistics are
	to be collected. Default is 1.
interval Specifies the number of seconds over which lmxdbstat
	summarizes activity.

(1) # lmxdbstat
1s port-rx buf-rx total-rx port-tx buf-tx total-tx total
pkt: 2 0 2 39 0 39 41
KB: 9 0 9 59 0 59 68 For example the first element in port-rx is
2:
2 = number of packets received in the last one second. The total number of packets transmitted and received is
total-rx/s + total-tx/s = 2 + 39 = 41 packets The second row indicates the number of Kbytes of data transmitted and received. The total amount of data transmitted/received in the last one second in Kbytes would be
9K + 59K = 68K

(2) #	lmxdbstat -p 21445

Showing LMX statistics for pid: 21445, Process: ora_lms1_rac10g1...
	1s pid ctx prx brx trx ptx btx ttx tot p: 21445 7 0 0 0 0 0 0 0 K: 21445 7 0 0 0 0 0 0 0

(3) #	lmxdbstat -d testdb

Showing LMX statistics for Database: testdb... Database Name: testdb Processes involved: ora_lmon_testdb1 (21410), ora_lmd0_testdb1 (21412), ora_lms0_testdb1 (21428), ora_lms1_testdb1 (21445) -------------------------------------------------------------- Snapshot time: 11-October-2007 15:04:19 -------------------------------------------------------------- prx brx trx ptx btx ttx tot utilization(%) 1p: 9 2005 2014 258 4 262 2276 100.0000 10p: 0 347 347 40 0 40 387 100.0000 30p: 0 119 119 13 0 13 132 100.0000 1K: 1 2843 2844 707 32 739 3584 100.0000 10K: 0 418 418 122 15 137 631 100.0000 30K: 0 169 143 42 19 61 231 100.0000 ============================================================================== The format of the above report is as follows:
	The first row shows database name. Second row shows Processes that are part of this database and also using LMX. Third row shows the time when the snapshot of LMX statistics was captured. This is important for the cases, where user wants to compare statistics of two or more databases.

Since the current implementation of lmxdbstat collects the LMX statistics from lmxstat and then compiles them for a database, total processing time for this might be more than interval second(s) and hence it may collect the statistics for more than interval second(s). This fact also necessitates timestamp (ie. Snapshot time) to be shown with each table for database statistics. Moreover, there are some transient Oracle process(es) apart from the instance specific processes that may use LMX and if while collecting the statistics for the database those process(es) disappear, then % utilization for that database may appear less than the actual utilization.

lmxstat(1M)