From c5a89386f273473954ec5f9e460052c175236c16 Mon Sep 17 00:00:00 2001 From: Thomas White Date: Thu, 15 Jul 2021 15:13:03 +0200 Subject: GUI: Run all jobs with same CWD as GUI itself This requires changing all the paths to relative ones. Previously, the jobs ran with their CWDs set to the corresponding "result" folder. The trouble is, this breaks everything that uses a relative location: image filenames, mask/satmap files referenced by the geometry, etc. Note that this change also means that the temporary folders ("indexamajig.XXX") are now produced in the top-level working directory. Eventually, a GUI option should be added to allow these folders to be put somewhere better (i.e. somewhere fast). Closes: https://gitlab.desy.de/thomas.white/crystfel/-/issues/44 --- src/gui_backend_slurm.c | 175 ++++++++++++++++++++++++++---------------------- 1 file changed, 95 insertions(+), 80 deletions(-) (limited to 'src/gui_backend_slurm.c') diff --git a/src/gui_backend_slurm.c b/src/gui_backend_slurm.c index e0f74b10..d07675be 100644 --- a/src/gui_backend_slurm.c +++ b/src/gui_backend_slurm.c @@ -172,13 +172,10 @@ static double indexing_progress(struct slurm_job *job, int *running) for ( i=0; in_blocks; i++ ) { char tmp[128]; - GFile *stderr_gfile; char *stderr_filename; snprintf(tmp, 127, "stderr-%i.log", i); - stderr_gfile = g_file_get_child(job->workdir, tmp); - stderr_filename = g_file_get_path(stderr_gfile); - g_object_unref(stderr_gfile); + stderr_filename = relative_to_cwd(job->workdir, tmp); n_proc += read_number_processed(stderr_filename); g_free(stderr_filename); @@ -469,8 +466,8 @@ static struct slurm_job *start_slurm_job(enum gui_job_type type, struct slurm_job *job; job_desc_msg_t job_desc_msg; submit_response_msg_t *resp; - GFile *stderr_gfile; int r; + GFile *cwd_gfile; script = load_entire_file(script_filename); if ( script == NULL ) return NULL; @@ -483,6 +480,8 @@ static struct slurm_job *start_slurm_job(enum gui_job_type type, job->type = type; + cwd_gfile = g_file_new_for_path("."); + env = create_env(&n_env, NULL); slurm_init_job_desc_msg(&job_desc_msg); @@ -499,7 +498,7 @@ static struct slurm_job *start_slurm_job(enum gui_job_type type, job_desc_msg.name = safe_strdup(jobname); job_desc_msg.std_err = strdup(stderr_filename); job_desc_msg.std_out = strdup(stdout_filename); - job_desc_msg.work_dir = g_file_get_path(workdir); + job_desc_msg.work_dir = g_file_get_path(cwd_gfile); job_desc_msg.script = script; job_desc_msg.environment = env; job_desc_msg.env_size = n_env; @@ -507,6 +506,8 @@ static struct slurm_job *start_slurm_job(enum gui_job_type type, job_desc_msg.account = safe_strdup(opts->account); job_desc_msg.array_inx = safe_strdup(array_inx); + g_object_unref(cwd_gfile); + r = slurm_submit_batch_job(&job_desc_msg, &resp); free(job_desc_msg.mail_user); free(job_desc_msg.partition); @@ -529,10 +530,7 @@ static struct slurm_job *start_slurm_job(enum gui_job_type type, job->job_id = resp->job_id; slurm_free_submit_response_response_msg(resp); - stderr_gfile = g_file_get_child(workdir, stderr_filename); - job->stderr_filename = g_file_get_path(stderr_gfile); - g_object_unref(stderr_gfile); - + job->stderr_filename = strdup(stderr_filename); job->workdir = g_file_dup(workdir); return job; @@ -560,11 +558,7 @@ static void write_partial_file_list(GFile *workdir, (i<(j+1)*block_size) && (iblock_size); - if ( !write_indexamajig_script(sc_filename, + sc_rel_filename = relative_to_cwd(workdir, "run_indexamajig.sh"); + files_rel_filename = relative_to_cwd(workdir, + "files-${SLURM_ARRAY_TASK_ID}.lst"); + stream_rel_filename = relative_to_cwd(workdir, + "crystfel-${SLURM_ARRAY_TASK_ID}.stream"); + stdout_rel_filename = relative_to_cwd(workdir, "stdout-%a.log"); + stderr_rel_filename = relative_to_cwd(workdir, "stderr-%a.log"); + harvest_rel_filename = relative_to_cwd(workdir, "parameters.json"); + + if ( !write_indexamajig_script(sc_rel_filename, proj->geom_filename, "`nproc`", - "files-${SLURM_ARRAY_TASK_ID}.lst", - "crystfel-${SLURM_ARRAY_TASK_ID}.stream", - serial_offs, 0, + files_rel_filename, + stream_rel_filename, + NULL, NULL, + harvest_rel_filename, + serial_offs, &proj->peak_search_params, &proj->indexing_params, wavelength_estimate, clen_estimate) ) { job = start_slurm_job(GUI_JOB_INDEXING, - sc_filename, + sc_rel_filename, job_title, array_inx, workdir, - "stdout-%a.log", - "stderr-%a.log", + stdout_rel_filename, + stderr_rel_filename, &opts->common); } else { job = NULL; } - g_free(sc_filename); if ( job != NULL ) { job->n_frames = proj->n_frames; @@ -675,6 +674,12 @@ static void *run_indexing(const char *job_title, } free(streams); + free(sc_rel_filename); + free(files_rel_filename); + free(stream_rel_filename); + free(stdout_rel_filename); + free(stderr_rel_filename); + free(harvest_rel_filename); g_object_unref(workdir); return job; @@ -856,37 +861,47 @@ static void *run_ambi(const char *job_title, struct slurm_job *job; struct slurm_ambi_opts *opts = opts_priv; GFile *workdir; - GFile *sc_gfile; - char *sc_filename; - GFile *stream_gfile; - char *stream_str; + char *sc_rel_filename; + char *stream_rel_filename; + char *stdout_rel_filename; + char *stderr_rel_filename; + char *fg_rel_filename; + char *intermediate_rel_filename; + char *harvest_rel_filename; workdir = make_job_folder(job_title, job_notes); if ( workdir == NULL ) return NULL; - stream_gfile = g_file_get_child(workdir, "ambi.stream"); - stream_str = g_file_get_path(stream_gfile); - g_object_unref(stream_gfile); - - sc_gfile = g_file_get_child(workdir, "run_ambigator.sh"); - sc_filename = g_file_get_path(sc_gfile); - g_object_unref(sc_gfile); - if ( sc_filename == NULL ) return NULL; - - if ( !write_ambigator_script(sc_filename, input, "`nproc`", - &proj->ambi_params, stream_str) ) + stream_rel_filename = relative_to_cwd(workdir, "ambi.stream"); + sc_rel_filename = relative_to_cwd(workdir, "run_ambigator.sh"); + stdout_rel_filename = relative_to_cwd(workdir, "stdout.log"); + stderr_rel_filename = relative_to_cwd(workdir, "stderr.log"); + fg_rel_filename = relative_to_cwd(workdir, "fg.dat"); + intermediate_rel_filename = relative_to_cwd(workdir, "ambigator-input.stream"); + harvest_rel_filename = relative_to_cwd(workdir, "parameters.json"); + + if ( !write_ambigator_script(sc_rel_filename, input, "`nproc`", + &proj->ambi_params, stream_rel_filename, + stdout_rel_filename, stderr_rel_filename, + fg_rel_filename, + intermediate_rel_filename, + harvest_rel_filename) ) { job = start_slurm_job(GUI_JOB_AMBIGATOR, - sc_filename, job_title, NULL, workdir, - "stdout.log", "stderr.log", &opts->common); + sc_rel_filename, job_title, NULL, workdir, + stdout_rel_filename, stderr_rel_filename, + &opts->common); job->niter = proj->ambi_params.niter; } else { job = NULL; } - g_free(sc_filename); + g_free(sc_rel_filename); + g_free(stdout_rel_filename); + g_free(stderr_rel_filename); + g_free(harvest_rel_filename); if ( job != NULL ) { - add_indexing_result(proj, job_title, &stream_str, 1); + add_indexing_result(proj, job_title, &stream_rel_filename, 1); } g_object_unref(workdir); @@ -903,19 +918,25 @@ static void *run_merging(const char *job_title, struct slurm_job *job; struct slurm_merging_opts *opts = opts_priv; GFile *workdir; - GFile *sc_gfile; - char *sc_filename; + char *sc_rel_filename; + char *output_rel_filename; + char *stdout_rel_filename; + char *stderr_rel_filename; + char *harvest_rel_filename; workdir = make_job_folder(job_title, job_notes); if ( workdir == NULL ) return NULL; - sc_gfile = g_file_get_child(workdir, "run_merge.sh"); - sc_filename = g_file_get_path(sc_gfile); - g_object_unref(sc_gfile); - if ( sc_filename == NULL ) return NULL; + sc_rel_filename = relative_to_cwd(workdir, "run_merge.sh"); + output_rel_filename = relative_to_cwd(workdir, "crystfel.hkl"); + stdout_rel_filename = relative_to_cwd(workdir, "stdout.log"); + stderr_rel_filename = relative_to_cwd(workdir, "stderr.log"); + harvest_rel_filename = relative_to_cwd(workdir, "parameters.json"); - if ( !write_merge_script(sc_filename, input, "`nproc`", - &proj->merging_params, "crystfel.hkl") ) + if ( !write_merge_script(sc_rel_filename, input, "`nproc`", + &proj->merging_params, output_rel_filename, + stdout_rel_filename, stderr_rel_filename, + harvest_rel_filename) ) { enum gui_job_type type; if ( strcmp(proj->merging_params.model, "process_hkl") == 0 ) { @@ -927,40 +948,34 @@ static void *run_merging(const char *job_title, } else { type = GUI_JOB_PARTIALATOR; } - job = start_slurm_job(type, sc_filename, job_title, NULL, - workdir, "stdout.log", "stderr.log", + job = start_slurm_job(type, sc_rel_filename, job_title, NULL, + workdir, stdout_rel_filename, + stderr_rel_filename, &opts->common); } else { job = NULL; } - g_free(sc_filename); if ( job != NULL ) { - GFile *hkl_gfile; - char *hkl; char *hkl1; char *hkl2; - hkl_gfile = g_file_get_child(workdir, "crystfel.hkl"); - hkl = g_file_get_path(hkl_gfile); - g_object_unref(hkl_gfile); - - hkl_gfile = g_file_get_child(workdir, "crystfel.hkl1"); - hkl1 = g_file_get_path(hkl_gfile); - g_object_unref(hkl_gfile); - - hkl_gfile = g_file_get_child(workdir, "crystfel.hkl2"); - hkl2 = g_file_get_path(hkl_gfile); - g_object_unref(hkl_gfile); + hkl1 = relative_to_cwd(workdir, "crystfel.hkl1"); + hkl2 = relative_to_cwd(workdir, "crystfel.hkl2"); - add_merge_result(proj, job_title, hkl, hkl1, hkl2); - g_free(hkl); + add_merge_result(proj, job_title, output_rel_filename, + hkl1, hkl2); g_free(hkl1); g_free(hkl2); } g_object_unref(workdir); + g_free(sc_rel_filename); + g_free(stdout_rel_filename); + g_free(stderr_rel_filename); + g_free(output_rel_filename); + g_free(harvest_rel_filename); return job; } -- cgit v1.2.3