Support

Alexander SPK NetWare FAQs
(Frequently Asked Questions)

What Versions of NetWare does the SPK support? The Alexander SPK supports NetWare 6, 5.x, 4.x and 3.x. It also supports NetWare 6.x and 5.x Multiprocessor; StandbyServer, and Cluster Services.

Is the SPK certified? The SPK has been tested and approved by both Novell and IBM. This includes certification under the Novell Yes Tested and Approved as well as the IBM ServerProven test program.

How will I know that a crash has occurred? If your server experiences a soft or hard crash, EDNA.NLM (Emergency Diagnostics for the Network Administrator) will display an Edna Crash Report Screen on the server. It will also generate SNMP traps that you can receive through both management and alert consoles. 

How will I know that a crash has occurred if the server is in another room or off-site? (See previous answer). 

My server has just experienced a crash. Now what do I do? First, from you workstation you need to open the EDNA.ECF located in sys:\edna\<date> where date is the date of the crash event in Y2K format. e.g. sys:\edna\20010629 for June 29, 2001. If there are additional crash events on the same day, the date will be followed by a letter. e.g. sys:\edna\20010629.a. Once the EDNA.ECF is open go to the button "Crash Cause". Approximately 70% of the time this screen will pinpoint the actual cause of the crash. This means you will often know who the culprit is right away without having to contact anyone else for assistance. Resolution of the remaining crash events can range from simple to complex. Many can be resolved after just a few minutes of examining the comprehensive Edna Crash File (ECF) under Windows further reducing the need to call for outside technical support. For the few challenging crashes that do require remote help, the SPK will create a self-extracting, emailable Ednamail.EXE file. Simply send it to Novell or another support provider for analysis. And remember that the recipient does not need to have the SPK since your ednamail.exe file includes the viewer.

Do Novell's technical support engineers analyze ECF files? Yes. Novell's technical support engineers analyze server crashes using the ECF files produced by the SPK.

How do I send Novell an Edna Crash File for analysis? When sending a crash file (ECF) to Novell for crash analysis, be sure to use the "Email" button on the EDNA.EXE Windows tool bar to build an email crash file. This option parcels together:
              Complete crash data
              Your contact information
              Any comments you care to add
              Provision for you to include an incident number
              The EDNA.EXE Windows crash file viewer
Edna then compresses it all into one self extracting file that you can attach to an email. All the support engineer needs to do at the other end is double click on the EDNAMAIL.EXE attachment and a complete crash file opens up immediately in front if him.
     NOTE: The EDNAMAIL.EXE can range in size from 5MB to 30MB. Some email servers limit the size of email accounts to 5MB or less causing the EDNAMAIL.EXE to not be sent or received properly. Sending the EDNAMAIL.EXE via an FTP server is an optional solution. You may also choose to put the EDNAMAIL.EXE on a zip disk or CD and send it though a parcel carrier. You may also want to rename the EDNAMAIL.EXE to COMPANYNAME.EXE so it can distinguished from other EDNAMAIL.EXE files sent for support.

What has to happen in order for the SPK to start working? Most of the time the SPK is quietly sitting in the background doing nothing. When NetWare declares an Abend (ABnormal END To Operations) the SPK intercepts the abend call and takes over the system. The SPK will then try to prevent the crash by suspending the entire offending NLM. If the SPK is successful in preventing the crash, thereby turning a hard crash into a soft crash, the Edna Crash Report Screen will be displayed on the server showing what was running when the event occurred.

What can I do if the server hangs without ever declaring an abend? Since the SPK is waiting for NetWare to declare an abend, we won't do anything. That's why we created a "backdoor" for such a situation. Welcome to the Debugger. With your right hand press Alt+Shift and with your left hand press Shift+Esc. Remember: Entering the debugger denies access to the server for all users. Since you are experiencing a system hang they probably can't get in anyway, so go on in. Once in the debugger type "eip=edna" <enter>. This will tell the processor to forget what he's currently doing, and do what you want him to do which is to go to Edna. Next type "g" for go <enter>. This will take you out of the debugger, create an Edna Crash File and RESTART the server. Once the server is back up and running, refer to the new crash file in the sys:\edna folder for information on what was causing the hang.

What can the SPK do that NetWare itself can't do? For the explanation of the differences between the two, click HERE.

How is the SPK priced? The SPK is sold on a per server basis. Pricing for SPK can be found HERE.

Can I use the SPK with a Standby Server? Yes. The SPK supports Standby Servers. When loading the SPK on StandbyServers, it is generally best to leave the “R” (Restart server after capturing hard crash) switch OFF. This enables StandbyServer to take over in the case of a primary server hard crash.

If I purchase a license for my primary server, will I have to purchase an additional license for my standby server? No. Since a standby server is only in use when the primary server goes down, you will not be violating the license agreement.

Can I use the SPK with  Cluster Servers? Yes. The SPK supports Cluster servers. When loading the SPK on Cluster Servers, it is generally best to leave the “R” switch ON and the “S” switch OFF. The “R” switch enables the SPK to restart the server after a hard crash. Without the SPK, NetWare Cluster Services requires human intervention to restart. The “S” (Soft crash if possible) switch allows operation of the failed module to be moved to an ancillary server in the cluster.

Do I need to purchase a license for every server in a cluster? Yes. Unlike a standby server that is only used in the event that the primary server goes down, all severs in a cluster are running concurrently. This requires that you purchase a license for each server in the cluster.