@GerryNelson  Thanks for this article which is essential when dealing with user directories or SCIM unable to provide POSIX attributes. The default format of the UID/GID displayed in the terminal by the identities plugin with text output mode can be misleading because , being a large integer value, it is displayed using scientific notation hiding the full digit value. I recommend switching to fulljson/json output mode in order to correct this behavior and display the untruncated integer value (sometimes up to 10, 12 digits long).  sas-viya --output fulljson identities show-group --id HR --show-advanced filtering out the result with jq might help also :-). ... View more

@LouGalway_sas  Thanks for sharing these important considerations which are, from a european-based perspective  especially relevant since accented letters (à la carte dishes, née Smith maiden name, piña colada cocktail recipe etc.) or diacritics are ubiquitous in non-english languages (German umlaut, French cedille, Spanish tildes, Romanian ă/î  to name only a few: every european written language might use such modified latin letters , some very commonly).  SAS treats differently accented letters whether you run an ASCII/ISO encoding session or a UTF8 session. The following  page explains it all : https://documentation.sas.com/doc/en/pgmsascdc/9.4_3.5/nlsref/p1pca7vwjjwucin178l8qddjn0gi.htmTherefore migrating data sets from ASCII charsets (Latin1, Latin9, WLATIN1 ) in addition to " affect[ing] data migration in terms of table sizes, data truncation, and the retention of indexes", also directly affects plain SAS code  : baseline syntax must be modified to take into account the presence of accented characters in CHAR variables, by removing accents altogether (if such alterations are acceptable) or replacing string functions with their k-counterparts.  This is very challenging indeed ! Perhaps, extending the native toolbox of SAS might help transitioning towards UTF8 :  providing an extended attribute to store in the descriptor portion of the Data Set the binary flag 1/0 (1= at least one CHAR var. with multi-bytes length) , sparing k-functions altogether (generalizing VARCHAR type with V9 engine might help greatly in this regard), providing NLS procs to detect in bulk the need for so-called internationalization (proc WhereIsMyMultiBytesVar data= ...) etc.    ... View more

@Vj_Govindarajan Thanks for your consideration and reformatting efforts : much easier to read now ! 🙂 I fully agree with @Patrick  : a SAS implementation would be a nice addition. ... View more

@Vj_Govindarajan Many thanks for this fundamental topic ! A cross-boundary approach was - still is, in many ways -  required to understand how SQL implementations differ for the same purpose. I came across the subject once or twice with Teradata and Snowflake when I was shown external SQL code which introduced the subject for me. A very seasoned team mate recently told me about it while refactoring SAS code. As mentioned, "Group By", "Remerging" are well-known expressions if and only if you know SAS already, since these concepts are peculiar to SAS and unintelligible for most other SQL users, whereas Window Functions are - by far in my opinion - more commonly known. Your article will therefore serve as reference on this subject and definitely help SAS as well as non SAS folks speak the same code, like a famous stone !  I have a small suggestion to make which might improve the readability of the text : could you systematically split up the SAS code inside the code blocks over multiple lines ? There are a couple of examples difficult to read on a single line. ... View more

@BrunoMueller Thank you for this correction ! I usually configure 'text' as default output mode and forces 'json' value specifically when needed. Now thanks to your tip, I'll use 'fulljson' as well :-). With 'fulljson' output, the activation status becomes indeed retrievable using the enabled (true/false) property. 🙂 ... View more

@RyanKing Thanks a lot for sharing this topic, too often overlooked, which is crucial in the real life. This post is very clear and engaging. Eagerly waiting for the next article ! ... View more

@BrunoMueller Thank you for this useful post which can serve as a perfect introduction to CLI rules handling on Viya ! Using the viya-cli authorization plugin (version 1.22.5) I was unable to get the rule _current_ enablement status : enabled = True/False. Oddly enough, the corresponding information cas be surfaced using the Rest API get/Rule operation whic provides the *enabled* field : https://developer.sas.com/rest-apis/authorization-v8/getRule Extracting the last activation/deactivation upon this rule from the audit records is feasible but tedious. ... View more

Thanks for this important clarification @RPoumarede , the new requirement is somewhat tiresome since it adds up to the CONNECT privileges still required (instead of replacing it altogether) but we are heading in the right direction. DBAs will remain our friends at the end of the day :-). ... View more

Hi @AnandVyas ,   @JuanS_OCS has provided you with the most comprehensive documentation you ought to know.   We use SAS Viya on OpenShift Bare Metal (on premises) with the following combination :   - Micro-service Stateful storage within ODF - SASWORK/CAS Cache directly accessing Worker Node local storage through HostPath - Shared Filesystem for SAS Studio through NAS/NFS to be replaced with GPFS.    Performance with ODF may vary, on our platform they use CEPH as backend storage behind ODF and ODF is going to be replaced by PortWorx in the coming months. No issue found with ODF so far but as you can see, critical storage consumed by SAS jobs , especially WORK, are not relying on ODF.   Ronan ... View more

Hi @GerryNelson ,   Thank you for this interesting post full of details. Is there any plan to integrate this new image into a standard k8s deployment part of Viya Manifest (optionnaly) installed alongside Viya microservices ? A single pod deployment would suffice.   The k8s cluster UI manager (Lens, Openshift console) would then provide a http *serverless* shell terminal to run VIYA Cli commands in addition to the ssh/docker/Jumpbox approach.   One step further would be to provide a new (http Restful) plugin for Web Environment Manager with a ssh console running *into* the Viya CLI Pod  under the authenticated user own credentials. Adding some additionnal CLI to the image (kubectl, kustomize, git, jq) would enable to replace the need for an extra SSH/Linux terminal with full security including client/server authentication sequence (cli auth login) automatically performed.   Ronan ... View more

Hi @GerryNelson ,   Thanks for this operation guide, detailed and well-researched. This is not an easy task, for sure ! At the beginning, in the commands given for finding out the primary pod controller, there is an undefined environment variable SOURCE_CLUSTER  :     export SOURCE_CLUSTER=sas-crunchy-platform-postgres PRIMARY=$(kubectl get pod --selector="postgres-operator.crunchydata.com/role=master,postgres-operator.crunchydata.com/cluster=$SOURCE_CLUSTER" -o jsonpath='{.items[*].metadata.name}' | awk '{print $1}') echo Primary Crunchy POD is $PRIMARY kubectl exec -it $PRIMARY -c database -- bash SOURCE_CLUSTER being defined  in the following page : https://documentation.sas.com/doc/fr/sasadmincdc/default/calsrvinf/n00000sasinfrdatasrv000admin.htm?fromDefault=#p12n56ik0ke0l4n18799bspg87l1   Instead of following the steep path of DB switching here described, another option could be to re-install Viya from scratch a-A (internal) => B-b (external) using the export/import feature or Environment Manager or  (better) SAS Viya CLI transfer plugin with additional steps as following : (a) export all Postgres content to JSON package , (b) import the previous package.   Best Regards, Ronan   ... View more

Thanks for the update @AlexBeaver  ! ... View more

Thanks for reposting here @AlexBeaver , the original pages at the  Scalability & Performance Focus Area RND site are now offline, unfortunately : https://support.sas.com/rnd/base/index.html 🤔 ... View more

Excellent post ! Thanks for sharing. I wasn't aware that fs-based shortcuts might be enabled when filesystem access is switched off : it works like an AllowList access very easy to configure. ... View more