How many zones does this ACS deployment have? Also, would it be possible for you to post some these duplicated entries for us here?
On Mon, Oct 2, 2017 at 8:40 AM, Richard Downer <rich...@apache.org> wrote: > Rafael, > > I have a sample `listTemplates` output. It's a real mixed bag. 5 templates > with crossZones==false have duplicated IDs, but duplicated a different > number of times. There are 4 instances of "GNU/Linux Fedora 20 - Minimal - > 64bits", 14 instances of "GNU/Linux Ubuntu 14.04 - Minimal - 64bits"; other > images duplicated 5 times or 8 times. > > If it is relevant - in each case, there are no other instances of the > template with other IDs (i.e. there are 4 "GNU/Linux Fedora 20 - Minimal - > 64bits" with a single ID, but no other templates with that name). > > Any theory on how this might have happened (database manual intervention)? > > If there's any other information you might find helpful please let me > know, although I do not have direct access to the ACS instance in question > (it's my client's client's ACS) so it may take me a little while to get > answers. > > Thanks for your help! > > Richard. > > > On 2017-10-02 11:31, Rafael Weingärtner <rafaelweingart...@gmail.com> > wrote: > > I would consider this a valid assumption. I checked one of the ACS > > environments (version 4.9.2, updated from 4.5) that I know use multiple > > zones, and all of the templates that have IDs duplicated have > > crossZones==true. > > > > We would need more information from this ACS that you are debugging to > pin > > point the problem. How many templates that have crossZones==false are > > presenting this duplication problem? If it is a bug, I would expect > > everyone of them to have duplicated IDs. However, if this is the only > thing > > they differ from other single zone templates, I would guess that these > > templates suffered from manual intervention in the database. > > > > On Mon, Oct 2, 2017 at 7:22 AM, Richard Downer <rich...@apache.org> > wrote: > > > > > Rafael, > > > > > > Thanks for the information. Is there any information in listTemplates > > > which would indicate if a template is in multiple zones? e.g. is the > > > `crossZones` parameter a reliable indicator - can we normally be > guaranteed > > > that there will be no duplicated IDs in templates where > crossZones==false? > > > > > > Currently jclouds will expect all templates where crossZones==false to > > > have unique IDs. I am trying to determine if this is a valid > assumption, or > > > if jclouds needs a bug fix to be prepared for a duplicate ID at any > time. > > > > > > Many thanks > > > Richard. > > > > > > > > > On 2017-09-29 15:24, Rafael Weingärtner <rafaelweingart...@gmail.com> > > > wrote: > > > > If the template is in multiple zones at the same time, then you are > going > > > > to have multiple template objects with the same ID. Otherwise, this > > > should > > > > not happen. You could check the database table "vm_template" to see > if > > > you > > > > find any inconsistency there. > > > > > > > > On Fri, Sep 29, 2017 at 9:21 AM, Richard Downer <rich...@apache.org> > > > wrote: > > > > > > > > > Hello, > > > > > > > > > > I'm debugging an issue on behalf of one of our users. We are using > > > Apache > > > > > jclouds to talk to Apache CloudStack which is on a customer's site. > > > > > > > > > > jclouds is invoking the `listTemplates` API, but is then choking > while > > > > > processing the query results. It seems that on this CloudStack > > > instance, > > > > > there are multiple templates with same ID - these all seem to > refer to > > > the > > > > > same OS image, but in different zones - and jclouds doesn't like it > > > when > > > > > things have the same ID but different content. > > > > > > > > > > Digging through the email archives, I found this message: > > > > > http://mail-archives.apache.org/mod_mbox/cloudstack-dev/ > > > > > 201501.mbox/%3c54c6c184.3010...@gmail.com%3E > > > > > ..which suggests that if a template is cross-zone, then it has the > > > same ID > > > > > in all zones. > > > > > > > > > > However the templates in this case all have "crossZones":false, > and the > > > > > above linked message sets the expectation that templates will NOT > have > > > the > > > > > same ID. > > > > > > > > > > Could somebody confirm the expectations for uniqueness of template > > > IDs? Is > > > > > it guaranteed that (under normal circumstances) template IDs are > unique > > > > > except for when crossZones is set - or is duplicate template IDs > > > always a > > > > > possibility? Does jclouds need to change its understanding of the > > > > > uniqueness of template IDs, or is there something odd about this > > > customer's > > > > > CloudStack installation? > > > > > > > > > > Thanks! > > > > > > > > > > Richard. > > > > > > > > > > > > > > > > > > > > > -- > > > > Rafael Weingärtner > > > > > > > > > > > > > > > -- > > Rafael Weingärtner > > > -- Rafael Weingärtner
