IBM TotalStorage SAN Switch Data Gathering
Abstract This TIP discusses IBM TotalStorage SAN Switch data gathering.
Contents SAN switch data gathering tips
The purpose of this TIP is to outline the minimum basic switch information that needs to be collected to assist in resolving SAN related problems. You might find it daunting to go through several steps to gather all the data requested, but the most common cause of delays in problem resolution is a lack of data. You should never assume that the root cause of any problem is in the most obvious place. By gathering logs from all parts of the SAN, you give yourself the greatest chance of getting a fast and effective resolution to the problem. The second most common cause of delays in problem resolution is providing data that has been collected some hours or even days after the problem occurred. Often in this case, there is no longer evidence of the original problem. Timely and complete data collection will aid in problems being resolved quickly. Collection of timely and detailed information for switches are outlined in the following sections. The collection of log information is critical to understanding the cause of an event in a SAN environment, and to aid support in analyzing the collected logs. It is also useful to provide the time offsets for individual equipment. As some hardware may have never had its real time clock set to the local time, it can become very difficult to match events from one piece of equipment to the other. Another important piece of information to aid in timely error analysis is a physical diagram of the SAN topology. This diagram should be kept up to date, and include all hosts, switches, directors, and storage devices within the SAN. This document can save countless hours of reconstructing the original picture by piecing together log information.
b-type family
Time difference Use the date command to display the switch date and time. Correcting the time This is a concurrent act and should be done if the displayed time is not accurate. The syntax is: Date "mmddHHMMyy" (month,day,hour,minute,year) Example: setting the time to 15:31 on Feb 27 2005 sw5:admin> date "0227153105" Thu Feb 27 15:31:00 2005 Log collection Using a telnet (or Secure telnet) client, log in to the switch and turn on session logging. At the command prompt issue the command: supportShow This needs to be performed for every switch in the fabric. After the supportshow completes, issue the commands: portLogClear This will clear the port activity logs portStatsClear x (where x is the port number you are interested in) The portstatsclear will clear the port statistics on every port attached to one ASIC, (normally a total of four ports). To clear the statistics for every port in the switch or director, the command must be specified for each quad within the switch. Hardware configuration collection A diagram showing how the SAN configuration is connected is of tremendous value.
Cisco family
Time difference From the CLI use the sh clock command to display the switch date and time.
Log collection CLI Using a telnet (or SSH) client, log in to the switch and turn on session logging. Issue the command: term len 0 (this will allow the output to continuously scroll) Issue the command: show tech-support details (or sh tech det for short) The term length can be set back to default by issuing term len 25 or it will be reset by logging off the CLI session. This should be performed for every switch in the fabric. Fabric Manager GUI The show tech might also be collected from the Fabric Manager menu by selecting the Tools —> collect techSupport menu. When collecting a show tech from the Fabric Manager interface, a JPG will be placed in the zip file that is the current fabric map view. To make the JPG meaningful, it is best to ensure the entire fabric is displayed in the map view and that it has been arranged for best viewing. Hardware configuration collection A diagram showing how your SAN configuration is connected is of tremendous value.
m-type family
Time difference From the hardware view of the switch in EFCM, go to the Configure menu and select the Date and Time menu, to get the switch date and time. Log collection From EFCM element manager window. Select Maintenance —> Data Collection. This procedure will create a zip file on your EFC Client machine, and we suggest saving the file to floppy disk. If the switches are not managed by EFCM, and SanPilot is used instead, then collect the output of the following commands in a telnet session to the switch. show switch show system show zoning show eventLog show features show loginServer show nameServerExt show port config show port info show port status show port technology show security fabricBinding show security portBinding show security switchBinding Tip: just entering show without a parameter, will put you into show mode which saves repeatedly typing show xxxxxx. Hardware configuration collection This will be included in the EFCM Data Collection procedure. However a configuration diagram showing how your SAN is connected is of tremendous value.
n-type family
Time Difference On the FC9000 Login to the FCM debug port using a null-modem cable. The current machine time is displayed at the top of the screen. If you set the Switch time using the GUI, the current time is displayed in the window. Note: UMD / 2045-N16, does not have a serial debug port. Log Collection A snapshot of the current machine state, including the logs is gathered by collecting a “DebugBackup” this can be collected from any client (even remotely), from the „File‟ pulldown and „Debug Backup‟. This client option will save the
resulting DebugBackup.zip to the Client workstation. Individual machines may be selected/deselected to only include the problem machine, and therefore keep the size of the resulting ZIP file to a minimum.