campaign issues

[pnorbert]

First access to campaign has debug prints, that should not be in master:

$ bpls -la ~/.campaign-store/test_s3d_31.aca
hostname: LAP131864.ornl.gov
-- Retrieved from DB data of mmd.0 size = 599 compressed = 1 compressed size = 599 original size = 1432 blob = 0x55d3aecf0568
-- Retrieved from DB data of md.idx size = 11045 compressed = 1 compressed size = 11045 original size = 1844320 blob = 0x55d3aecf0df8
-- Retrieved from DB data of md.0 size = 27249181 compressed = 1 compressed size = 27249181 original size = 248832248 blob = 0x7fe0f629f018
-- Retrieved from DB data of profiling.json size = 303219 compressed = 1 compressed size = 303219 original size = 3458742 blob = 0x55d3aeef09a8
Errno was 111
  double   ../data/ptj.field.bp/pressure  31*{2560, 960, 3456} = 1.79769e+308 / 2.22507e-308
  double   ../data/ptj.field.bp/species   31*{19, 2560, 960, 3456} = 1.79769e+308 / 2.22507e-308
  double   ../data/ptj.field.bp/temp      31*{2560, 960, 3456} = 1.79769e+308 / 2.22507e-308
  double   ../data/ptj.field.bp/uvel      31*{2560, 960, 3456} = 1.79769e+308 / 2.22507e-308
  double   ../data/ptj.field.bp/vvel      31*{2560, 960, 3456} = 1.79769e+308 / 2.22507e-308
  double   ../data/ptj.field.bp/wvel      31*{2560, 960, 3456} = 1.79769e+308 / 2.22507e-308

There is also a strange Errno 111 error printed, but it's not obvious there is an error.

adios@LAP131864:~/test/demo.uk_meeting$ bpls -la ~/.campaign-store/test_s3d_31.aca
hostname: LAP131864.ornl.gov
Errno was 111
  double   ../data/ptj.field.bp/pressure  31*{2560, 960, 3456} = 1.79769e+308 / 2.22507e-308
  double   ../data/ptj.field.bp/species   31*{19, 2560, 960, 3456} = 1.79769e+308 / 2.22507e-308
  double   ../data/ptj.field.bp/temp      31*{2560, 960, 3456} = 1.79769e+308 / 2.22507e-308
  double   ../data/ptj.field.bp/uvel      31*{2560, 960, 3456} = 1.79769e+308 / 2.22507e-308
  double   ../data/ptj.field.bp/vvel      31*{2560, 960, 3456} = 1.79769e+308 / 2.22507e-308
  double   ../data/ptj.field.bp/wvel      31*{2560, 960, 3456} = 1.79769e+308 / 2.22507e-308

Third,
adios2_campaign_manager.py has too many arguments, -p, -a, -s should be just a single full name. Support for full path to a campaign file would also be nicer, to work like bpls in the above example.

[pnorbert]

PR #4048 can be added to release 2.10 only if the first two issues are fixed.

[pnorbert]

The second issue about Errno 111 is that the remote server is cannot be connected (I forgot to launch it). We need a good error message here

[dmitry-ganyushin]

The second issue about Errno 111 is that the remote server is cannot be connected (I forgot to launch it). We need a good error message here

This comes from evpath

	printf("Errno was %d\n", errno);

but the evpath library has very limited support.

[eisenhauer]

The nice thing about EVPath is that it can have all the support we need it to have... Yeah, the printf() needs to be removed and the lack of a connection caught at the ADIOS2 level. The errno printout snuck in there when I was debugging something relatively recently. The not throwing an error is the real bug. Will sort that soon. I'm not sure we can get (or need) a better error than "connection to remote server failed". About the only other thing we might detect with more support from EVPath would be if you specified a hostname that couldn't be translated via DNS or something like that.

[dmitry-ganyushin]

The following code is pretty common:

void Remote::Open(const std::string hostname, const int32_t port, const std::string filename,
                  const Mode mode, bool RowMajorOrdering)
{

    m_conn = CMinitiate_conn(ev_state.cm, contact_list);
    if (!m_conn)
        return;

The result is the same in both cases if connection is successful and not successful.

[eisenhauer]

Well, not quite the same if you look at more than the first two lines... If we complete the Open() the m_Active member (which defaults to false) is set to true and the Remote's boolean operator returns the value of m_Active, so we know when the open succeeds. The boolean operator is used later to know if we can/should use the remote connection.