Using ACS 1.1.8.3, when opening open source pkg mgmt ....

[Michael Mayer]

the RPM packages I’ve downloaded and installed, most are missing from the list. *SSHD is started on the IBM i. When trying to install a package as a test, we see an “rpmdb changed underneath us” error. If we try to import a package, we see the screen shot below.

We though this might be a firewall error but this looks like it could be a different issue. IBM says it’s something they don’t deal with.

They said it is on the GIT / Bit Bucket side.

Help please.

‌

[Michael Mayer]

Here’s what we get if we try to reinstall a package,

‌

[Michael Mayer]

I had already installed bootstrap several months ago when weI went from 5733OPS to Open source via ACS. We didn’t have any issues. All packages downloaded and installed without error. Then yesterday, out of the blue, almost 300 pkgs we had installed, only 34 were displayed. After running the IBM SQL bootstrap installer (shown below) again and having it fail: due to QOpenSys/pkgs/bin/bash already existing, we received the message to add “-f” to the ‘bin’bash' file to remove it. I didn’t want to do that since I didn’t know what the implications were. Ends up the line of code to remove ‘/QOpenSys/pkgs/bin/bash’ can be added to the ‘bootstrap.sh’ file in WRKLNK ‘/tmp’ folder. In another post by Andrew Borts, (who I’ve followed for a long time), he mentioned something about adding an environment variable on a similar problem. ADDENVVAR ENVVAR(QIBM_MULTI_THREADED) VALUE(Y) - Environment variable added. I ran that off a normal 5250 command line.

Next, we closed out ACS, endtcpsvr *sshd, strtcpsvr *sshd, opened ACS and the Open Source Package Mgmt and now there are 51 yum packages shown. We just installed the other 247 packages without issue.

‌

create or replace table qtemp.ftpcmd(cmd char(240)) on replace delete rows;
create or replace table qtemp.ftplog(line char(240)) on replace delete rows;

insert into qtemp.ftpcmd(CMD) values 
   ('anonymous [e-mail]')
  ,('namefmt 1')
  ,('lcd /tmp')
  ,('cd /software/ibmi/products/pase/rpms')
  ,('bin')
  ,('get README.md (replace')
  ,('get bootstrap.tar.Z (replace')
  ,('get bootstrap.sh (replace')
  with nc
;

CL:OVRDBF FILE(INPUT) TOFILE(QTEMP/FTPCMD) MBR(*FIRST) OVRSCOPE(*JOB);
CL:OVRDBF FILE(OUTPUT) TOFILE(QTEMP/FTPLOG) MBR(*FIRST) OVRSCOPE(*JOB);

CL:FTP RMTSYS('public.dhe.ibm.com');

CL:QSH CMD('touch -C 819 /tmp/bootstrap.log; /QOpenSys/usr/bin/ksh /tmp/bootstrap.sh > /tmp/bootstrap.log 2>&1');

select
case when (message_tokens = X'00000000')
 then 'Bootstrapping successful! Review /tmp/README.md for more info'
 else 'Bootstrapping failed. Consult /tmp/bootstrap.log for more info'
end as result
from table(qsys2.joblog_info('*')) x
where message_id = 'QSH0005'
order by message_timestamp desc
fetch first 1 rows only;

‌

[Kevin Adler]

Why did you re-run the bootstrap?

[Michael Mayer]

Kevin Adler commented on issue #107:
Using ACS 1.1.8.3, when opening open source pkg mgmt .... Why did you re-run the bootstrap?

‌

M. Mayer replied on issue #107:

We figured it couldn’t hurt. Then I noticed on 1 of the issues where you said it only needs to be run once. No harm, no foul. What was really the issue with the open source packages that we installed several months back not showing in ACS and having to reinstall them?

‌

[Kevin Adler]

Re-running the bootstrap will replace your rpm database with the one in the bootstrap. This means that rpm and yum will have no idea what software is already installed. In addition, any software in the bootstrap which you have upgraded in the mean time will be back-leveled to the version in the bootstrap. This process doesn’t hurt per se, but it certainly would have done nothing to solve your original problem and in fact only exacerbated things.

‌

Unfortunately to solve your original problem, all the necessary information has been lost due to re-applying the bootstrap. I don’t think we can solve that one any more.

‌

[Michael Mayer]

In the final analysis, are we ok? If another firm were to re-run the bootstrap, what, if anything, should be deleted before hand? I haven’t worked with open source and am an admin, not a pgmr, but I wanted to tool around with it to get familiar with it.

[Michael Mayer]

I see. In the final analysis, are we ok? If someone were to re-run the bootstrap like we did, are there steps that should be taken before doing it? If so, what? Maybe delete ….. ?

‌

[Kevin Adler]

You should be ok, especially if you re-installed all the software that was previously installed. As re-running the bootstrap is not something we support, I don’t think we would document what needs to be done to re-run it.

‌

We will however update the bootstrap to prevent re-running the bootstrap on a system that has already been bootstrapped.

[Michael Mayer]

Thank you. Much appreciated.

‌

[Michael Mayer]

It would be interesting to know if other customers ran into the same situation where the RPM’s just disappeared after a short period of time for some unknown reason.

[Jesse G]

We have opened an internal issue to track if other customers hit similar problems to the original report. Thanks for opening this issue!