Ben, You said something that I would like clarification on:
any attributes (of our hand-constructed classes using default Admin defined field ids) (italics added) Just to be certain I’m understanding things correctly, did you create these classes using the CMDB console using the functionality there for creating classes and adding attributes to them, or did you do any of this work in the Admin tool? While I haven’t added any new classes, I have added a fair number of attributes to existing classes (although we used IDs specified from a specific range of IDs that we were using, rather than letting the system auto-assign IDs), and we never had a problem with the system not bringing these attributes into the sandbox, which is something that I think should be someone analogous to what you’re trying to do in the end. While I’m not an expert on what happens when you use a sample schema, I would expect that the fact that “by ID” is selected, it should bring across all fields with matching IDs regardless of whether or not they’re in the sample schema – which it seem would have to be the case or the majority of the OOB classes would have this same issue. Well, I guess there’s no useful information in the paragraph above. I guess I really just want to confirm that the classes were all created using the CMDB console exclusively, and that any AST forms used to work with those classes were created using the CMDB2Asset utility so that the IDs of the fields on the asset forms and the CMDB forms all match up, regardless of whether they’re OOB or custom. Lyle From: Action Request System discussion list(ARSList) [mailto:[email protected]] On Behalf Of Ben Chernys Sent: Wednesday, January 28, 2009 1:57 PM To: [email protected] Subject: Re: ITSM Issue with Sandbox Enablement ** Thanks Guys, I agree with you. I noticed the part that doesn't make sense but believe it or not I have a reasonable reason for using it. As for auditing, we've had our share of issues with it but are (over) using it. I disabled the delete activity in the Sandbox reconciliation job so that when a CI in Production (Gold?) is touched by a person, I can have our standard reconciliation job basically use the equal Sandbox CI at a higher precedence and then more or less cancel the normal discovered CI in the recon job. When that normal discovered CI has "caught up" to the human modified CI, in a set of configured fields I might add, then the Sandbox CI is deleted and the CI is no longer human modified and participates in reconciliation updates as normal. I also cannot add an attribute indicating the human modification state to BaseElement. The Sandbox indeed makes no sense whatsoever as implemented with the OOTB workflow. The Updated CI is pushed (supposedly) to the Sandbox, and then immediately reconciled to production AND deleted from Sandbox. Also note the comments about the NULL Option in the OOTB Job. This may be there to cover the issue I encountered below. We now have a ticket with BMC but alas, I am going to have to do something to make it work sooner than I expect a response or a fix. Perhaps the Overlay feature works better? Cheers Ben ________________________________ From: Action Request System discussion list(ARSList) [mailto:[email protected]] On Behalf Of Guillaume Rheault Sent: January 28, 2009 9:15 PM To: [email protected] Subject: Re: ITSM Issue with Sandbox Enablement ** I totally agree with Lyle. It just does not make sense to have a sandbox that will be immediately reconciled to production dataset. Besides restricting the updates via permissions or field properties, I would add that enabling auditing on the fields that can be updated is a good complement, from a control/audit perspective. So with access control and auditing, you really don't need a sandbox, and you are going to gain a lot in terms of reducing complexity, maintenance, etc. -Guillaume ________________________________ From: Action Request System discussion list(ARSList) on behalf of Lyle Taylor Sent: Wed 01/28/09 2:52 PM To: [email protected] Subject: Re: ITSM Issue with Sandbox Enablement In my last position we struggled and fought with the Sandbox for quite a while, finding issues here and there, and finally came to the conclusion that the Sandbox provides no real value as it is currently implemented (at least ITSM pre-7.5, I haven’t see how they’ve changed it there yet), is fatally flawed in a few different ways and just caused more headaches than it was worth. In the end, we decided to simply turn it off, and I would recommend the same unless you have a specific need for it. In my opinion, the only real and valid benefit that could be obtained from a Sandbox would be as a staging area to make changes that are to be applied to the Gold dataset at a later time. The Sandbox implemented in CMDB 2.1 and earlier does not work this way. It is merely a pass-through dataset that gets immediately reconciled into Gold. As such, I see no benefit in using it. You could argue that it gives you the ability to limit what kinds of changes a person can make to Gold by changing the reconciliation rules. However, I would argue that if you don’t want someone to change something, because you have a better (automated) source for the information, a more appropriate solution would be to simply not let them change it in the first place via an active link or something that makes the field read only. It is extremely poor user interaction design to let someone make a change and save it, only to wonder why it didn’t take effect afterward. I know this isn’t what you were looking for, but after everything I went through trying to correctly use the Sandbox, I would recommend to just about anyone to leave it be and turn it off until its design and implementation get fixed. Good luck, Lyle From: Action Request System discussion list(ARSList) [mailto:[email protected]] On Behalf Of Ben Chernys Sent: Wednesday, January 28, 2009 2:26 AM To: [email protected] Subject: FW: ITSM Issue with Sandbox Enablement ** Hi Folks I am having a ticket raised with BMC for this one but I just thought I'd pass it to the list for any thoughts that may come my way. It has me in a bit of a quandary and I thank anyone that can help me resolve it or work-around it. I have placed the log zip file (88KB) here because the list software blocks my attachment. www.softwaretoolhouse.com/_logs/ARUSERC2.ZIP<http://www.softwaretoolhouse.com/_logs/ARUSERC2.ZIP> Platform Sparc Sun Fire V240 OS Solaris 5.10 DB Oracle 10g2 ARS 7.1 patch 5 ITSM 7.0.3 patch 7 Cheers Ben Chernys Senior Software Architect Software Tool House Inc. Canada / Deutschland / Germany Mobile: +49 171 380 2329 GMT + 1 + [ DST ] Email: mailto:[email protected] Web: http://www.softwaretoolhouse.com<http://www.softwaretoolhouse.com/> A free notepad for Diary fields: http://www.softwaretoolhouse.com/downloads/DiaryFieldEditor.htm An ARS API scripting tool used for migrations, integrations, imports, reports, extracts, batch jobs: http://www.softwaretoolhouse.com/products/SthMupd ______________________________________________ Von: Chernys, Ben Gesendet: Mittwoch, 28. Januar 2009 09:37 An: Ben Chernys Cc: Betreff: ITSM Issue with Sandbox Enablement When we use the Sandbox feature (btw I had to put a fix in for our Geman clients - Sandbox enablement is ignored if you are not running an English client - any attributes (of our hand-constructed classes using default Admin defined field ids) that are NOT in BaseElement (or rather BMC.CORE:BMC_MainFrame) do NOT get pushed to the Sandbox instance. The Sandbox job has NULL Defer = Yes in the OOTB job. I can see why. If you turn this off (which is a bug that the OOTB is NOT off - restricting you from nullifying an attribute and then causing a mis-match between the two instances in the datasets) what happens is all the non-BaseElement (MF) attributes become NULL even when they are not touched. I believe the error is manifested by the filter ASI:SHR:All_600_PushToBMCForm which uses a sample schema where the push target is in a DO field. The filter has the "by ID" check-box checked. The Log describes the fields pushed and those target fields include those fields for the real target schema but the values for these fields are all null. Have you seen this behaviour? You should notice it if you take any class which other than BMC_Mainframe and change an attribute from that class (which field id is NOT in BMC_Mainframe). The problem is isolated to retrieving the values of fields as the target fields seem to be complete. I have checked the database (filter_push) and will need to have a play with the API to see what actually is set for a "by like id"push fields. It is possible, I suppose, to build a better sample form with all of our and OOTB field ids in it. from filter "ASI:SHR:All_600_PushToBMCForm", 3093, 1226599817, "Remedy", "panacea", 1, 600, 20, 1, 1, 2, "ZODP+HGF8UUQFMpti/TK3KBO9J67T2saQn68e9TkISfKv8K219ABUBhboLdYUUv0UBd00rC2s98yWtiDld8iwnwpzvEXvjEb", "4\6\99\301170700\2\0", NULL, "1144618550♦BMC♦Copyright (c) 1991 - 2006 BMC Software, Inc. all rights reserved BMCVer=7.00.00♥", NULL, "4\60006\4\0\\60008\40\0\60009\4\0\\60010\4\0\\", NULL, NULL, 0, 0 from filter_push 3093, 0, 98, "1...@\11\$301170700$\1\98\4\1\1\179\99\179\4\5\102\1\@\...@\1\98\0\4\5\", NULL, "BMC.CORE:BMC_Mainframe", "@" 3093, 1001, 98, "1...@\11\$301170700$\1\98\4\1\1\179\99\179\4\3\102\1\@\...@\1\98\0\4\3\", NULL, "BMC.CORE:BMC_Mainframe", "@" As far as I can tell there will be two work-arounds possible: 1) a better sample form. or 2) a filter for each class replacing the single filter above. The conclusion or direction has changed since I implemented the following test. I replaced the above filter with one using the exact form that was participating in the Push Fields. Same effect. It now looks that this SandboxCreate CI Name causes untraced actions in the hiddent Invoke External Filter CMDB Processes and that it is likely that the error is there. I have attached a client trace file (AL, Filter, SQL, API). We have turned off (deactivated) any non-OOTB filters. The ASI:SHR:All_600_PushToBMCForm has been changed to specify the same form as the value of the OS Schema field in the trace. 093411.591 i ArQryGet returns 1 records for select name , queryshort from filter where queryshort like '%Reconcile%' "001""002" <-------------------->SQL row: 1 Col 0: ASI:SHR:SandboxCallReconEngineRelation_999 Col 1: 4\1\99\1000000076\2\4\9\Reconcile\ 093422.294 i ArQryGet returns 2 records for select name , queryshort from filter where queryshort like '%Reconclie%' "001""002" <-------------------->SQL row: 1 Col 0: ASI:SHR:SandboxCallReconEngine_999 Col 1: 1\4\1\99\400127400\99\1714700\4\1\99\301774800\2\4\9\Reconclie\ <-------------------->SQL row: 2 Col 0: ASI:ALK:CheckQty_450 Col 1: 1\1\2\4\5\99\301149300\2\2\0\4\1\99\301149300\2\0\4\6\99\301774800\2\4\9\Reconclie\4\6\50\200000020\2\4\13\SandboxCreate\ Have you seen this behaviour? And can you recommend anything to get around this? This is a show-stopper for us for now. Cheers Ben Chernys <<ARUSERC2.ZIP>> . __Platinum Sponsor: RMI Solutions ARSlist: "Where the Answers Are" html___ NOTICE: This email message is for the sole use of the intended recipient(s) and may contain confidential and privileged information. Any unauthorized review, use, disclosure or distribution is prohibited. If you are not the intended recipient, please contact the sender by reply email and destroy all copies of the original message. __Platinum Sponsor: RMI Solutions ARSlist: "Where the Answers Are" html___ __Platinum Sponsor: RMI Solutions ARSlist: "Where the Answers Are" html___
