swa report returns an error

[Steven Haywood (SA)]

# swa report

======= 02/02/07 13:45:01 SASTT BEGIN Report on Issues and New Software (user=utgsxh) (jobid=git5.pta.za.go2uti.com)

* Gathering Inventory
* Checking existence and age of inventory for host "git5.pta.za.go2uti.com"
* Inventory for host "git5.pta.za.go2uti.com" does not exist
* Listing Filesets
* Listing Products
* Listing Bundles
* Inventory written to /roothome/.swa/cache/swa_inventory_2106257851.xml
* Getting Catalog of Recommended Actions and Software
* Checking existence and age of local catalog file
* Local catalog file does not exist
* Downloading "/roothome/.swa/cache/swa_catalog.xml" from the remote location "https://ftp.itrc.hp.com/wpsl/bin/doc.pl/screen=wpsl
DownloadPatch/swa_catalog.xml.gz?PatchName=/export/patches/swa_catalog.xml.gz"
* Uncompressing stream
* Download: 4kB read
* Download: 1323kB read
* Download: 2035kB read
* Download: 3178kB read
* Download: 3448kB read
* Download: 3584kB read
* Download: 4038kB read
* Download: 4767kB read
* Download: 5042kB read
* Download: 5394kB read
* Download: 5664kB read
* Download: 6074kB read
* Download: 6279kB read
* Download: 6799kB read
* Download: 7983kB read
* Download complete: 8491kB
* Successfully downloaded catalog
* Performing Analysis
ERROR: Inventory file: /roothome/.swa/cache/swa_inventory_2106257851_1.xml does not exist.

======= 02/02/07 13:49:54 SASTT END Report on Issues and New Software failed with 1 error. (user=utgsxh)
(jobid=git5.pta.za.go2uti.com)

Help please, I'm an HPUX noob :)

Thanks
Steven

[Steven Haywood (SA)]

Scratch the question - I got it working by using swa report -s git5.pta.za.go2uti.com
No idea why, but it's all good now :)

[Bob E Campbell · Accepted solution]

The error message is cryptic I agree. The issue here is that the SWA tool was meant to allow a single system to analyze and maintain other remote systems. To support this a cache of system inventories is maintained within the ~/.swa/cache directory. It appears that the tool believed that there was a collision of some kind (thus the _1.xml). By specifying a full DNS name for the target the cache treated this as a new system (not very smart yet).

This may have come about if you had multiple analysis runs in parallel. If you need to do this use the "-x user_dir" option to give each command its own user cache.

You should be able to reset these files using the command "swa clean usercache". If you reproduce the problem please let me know the steps taken.

Bob

[Olivier Masse]

Similar problem here. I delete /root/.swa completely, then run swa report and get this:

======= 03/12/07 11:55:18 EDT BEGIN Report on Issues and New Software
(user=ch7558) (jobid=paiutlux01)

* Gathering Inventory
* Checking existence and age of inventory for host "paiutlux01"
* Inventory for host "paiutlux01" does not exist
* Listing Filesets
* Listing Products
* Listing Bundles
* Inventory written to /root/.swa/cache/swa_inventory_316520417.xml
* Getting Catalog of Recommended Actions and Software
* Checking existence and age of local catalog file
* Local catalog file does not exist
* Downloading "/root/.swa/cache/swa_catalog.xml" from the remote
location "https://ftp.itrc.hp.com/wpsl/bin/doc.pl/screen=wpslDownloadP
atch/swa_catalog.xml.gz?PatchName=/export/patches/swa_catalog.xml.gz"
* Uncompressing stream
* Download complete: 8495kB
* Successfully downloaded catalog
* Performing Analysis
ERROR: Inventory file: /root/.swa/cache/swa_inventory_316520417_1.xml does
not exist.

======= 03/12/07 11:55:38 EDT END Report on Issues and New Software failed
with 1 error. (user=ch7558) (jobid=paiutlux01)


It does indeed create the .xml file, but later it seems to try to access _1 which, obviously, does not exist.

If I run it again, I get the same error but with _2.

[Steven Haywood (SA)]

Hi
The workaround I used was to use swa report -s

Give it a shot and see if it works?

Thanks
Steven

[Olivier Masse]

Yup, that worked, thanks.

I think the cause is that my hostname is different from the nodename (as reported by uname -n) since our corporate standards require hostnames that are longer than 8 characters. Never had any problem until now with this.

Can't hand out points on this one since you're the original poster, sorry.

[Steven Haywood (SA)]

LOL no worries - glad you came right :)

[Bob E Campbell]

The cause of this problem turns out to be that older and newer parts of SWA differed on whether the default should be the nodename or hostname. If the returned values from "hostname" and "uname -n" do not match you will have this problem.

By setting an explicit value you avoid this difference in the defaults.