Merge pull request #305 from NERSC/fix-wrapper

Fix logging and update for Slurm

.travis.yml

@@ -1,5 +1,5 @@
 sudo: true
-dist: trusty
+dist: bionic
 language: python
 python:
   - "2.7"
@@ -9,9 +9,9 @@ compiler:
 # command to install dependencies
 before_install:
   - sudo chmod 755 /var/log
-  - sudo add-apt-repository "deb http://us.archive.ubuntu.com/ubuntu/ precise universe multiverse"
+  - sudo add-apt-repository "deb http://us.archive.ubuntu.com/ubuntu/ bionic universe multiverse"
   - sudo apt-get update -qq
-  - sudo apt-get install -qq libjson-c2 libjson-c-dev munge libmunge2 libmunge-dev libcurl4-openssl-dev autoconf automake libtool curl make valgrind xfsprogs squashfs-tools libcap-dev
+  - sudo apt-get install -qq libjson-c-dev munge libmunge2 libmunge-dev libcurl4-openssl-dev autoconf automake libtool curl make valgrind xfsprogs squashfs-tools libcap-dev
   - sudo modprobe squashfs
   - sudo modprobe xfs
   - cat /proc/filesystems
@@ -28,8 +28,9 @@ before_script:
   - cp imagegw/test.json.example test.json
   - mkdir /tmp/imagegw /tmp/systema /tmp/systemb
   - rm imagegw/test/sha256sum
-  - find /tmp/
 script:
+  - echo "Disabling autoclear for Travis and Kernel 5.4"
+  - sed -i 's/useAutoclear = 1;/useAutoclear = 0;/' ./src/shifter_core.c
   - export PROC_COUNT=$(cat /proc/cpuinfo  | egrep '^processor' | wc -l)
   - ./autogen.sh
   - ./configure --prefix=/usr --sysconfdir=/etc/shifter --disable-staticsshd

extra/CI/integration_test.sh

@@ -46,7 +46,7 @@ done
 
 echo "setting up munge"
 if [[ -e /usr/sbin/create-munge-key ]]; then
-    sudo /usr/sbin/create-munge-key
+    sudo /usr/sbin/create-munge-key -f
 fi
 sudo service munge start
 

shifter.spec

@@ -19,7 +19,7 @@
 
 Summary:   NERSC Shifter -- Containers for HPC
 Name:      shifter
-Version:   18.03.5
+Version:   18.03.6
 Release:   %{shifter_release}
 License:   BSD (LBNL-modified)
 Group:     System Environment/Base

wlm_integration/slurm/shifterSpank.c

@@ -330,7 +330,7 @@ int forkAndExecvLogToSlurm(const char *appname, char **args) {
             if (stdoutStream) {
                 ssize_t nBytes = getline(&lineBuffer, &lineBuffer_sz, stdoutStream);
                 if (nBytes > 0) {
-                    _log(LOG_ERROR, "%s stdout: %s", appname, lineBuffer);
+                    _log(LOG_DEBUG, "%s stdout: %s", appname, lineBuffer);
                 } else {
                     fclose(stdoutStream);
                     stdoutStream = NULL;
@@ -348,7 +348,7 @@ int forkAndExecvLogToSlurm(const char *appname, char **args) {
         }
 
         /* wait on the child */
-        _log(LOG_ERROR, "waiting on %s\n", appname);
+        _log(LOG_INFO, "waiting on %s\n", appname);
         waitpid(pid, &status, 0);
         if (WIFEXITED(status)) {
              rc = WEXITSTATUS(status);
@@ -928,7 +928,7 @@ int shifterSpank_job_prolog(shifterSpank_config *ssconfig) {
         set_type = 1;
     }
 
-    _log(LOG_ERROR, "about to lookup image in prolog env");
+    _log(LOG_INFO, "about to lookup image in prolog env");
     ptr = getenv("SPANK_SHIFTER_IMAGE");
     if (ptr != NULL) {
         char *tmp = imageDesc_filterString(ptr, set_type ? ssconfig->imageType : NULL);
@@ -1144,7 +1144,7 @@ int shifterSpank_job_prolog(shifterSpank_config *ssconfig) {
     strncpy_StringArray(ssconfig->image, strlen(ssconfig->image), &setupRootArgs_sv, &setupRootArgs, &n_setupRootArgs, 10);
 
     for (setupRootArgs_sv = setupRootArgs; setupRootArgs_sv && *setupRootArgs_sv; setupRootArgs_sv++) {
-        _log(LOG_ERROR, "setupRoot arg %d: %s", (int)(setupRootArgs_sv - setupRootArgs), *setupRootArgs_sv);
+        _log(LOG_INFO, "setupRoot arg %d: %s", (int)(setupRootArgs_sv - setupRootArgs), *setupRootArgs_sv);
     }
 
     /* setup memory cgroup if necessary */
@@ -1168,7 +1168,7 @@ int shifterSpank_job_prolog(shifterSpank_config *ssconfig) {
 
     int status = forkAndExecvLogToSlurm("setupRoot", setupRootArgs);
 
-    _log(LOG_ERROR, "after setupRoot, exit code: %d", status);
+    _log(LOG_INFO, "after setupRoot, exit code: %d", status);
 
 
     if (status == 0) {
@@ -1183,7 +1183,7 @@ int shifterSpank_job_prolog(shifterSpank_config *ssconfig) {
     }
 
     pid = findSshd();
-    _log(LOG_DEBUG, "shifter_prolog: sshd on pid %d\n", pid);
+    _log(LOG_INFO, "shifter_prolog: sshd on pid %d\n", pid);
 
 _prolog_exit_unclean:
     if (setupRootArgs != NULL) {
@@ -1289,7 +1289,7 @@ int shifterSpank_job_epilog(shifterSpank_config *ssconfig) {
         rc = SLURM_ERROR;
     }
 
-    _log(LOG_DEBUG, "shifter_epilog: done with unsetupRoot");
+    _log(LOG_INFO, "shifter_epilog: done with unsetupRoot");
 
 _epilog_exit_unclean:
     return rc;

wlm_integration/slurm/wrapper.c

@@ -331,18 +331,21 @@ int setup_libslurm() {
     return ERROR;
 }
 
-extern int slurm_stepd_connect(const char *dir, const char *node, uint32_t job, uint32_t step, uint16_t *protocol);
+extern int slurm_stepd_connect(const char *dir, const char *node, slurm_step_id_t *step, uint16_t *protocol);
 extern int slurm_stepd_add_extern_pid(uint32_t, uint16_t, pid_t);
 extern hostlist_t slurm_hostlist_create_dims(const char *hostlist, int dims);
 extern char * slurm_hostlist_deranged_string_malloc(hostlist_t hl);
 
 int wrap_spank_stepd_connect(shifterSpank_config *ssconfig, char *dir,
     char *hostname, uint32_t jobid, uint32_t stepid, uint16_t *protocol)
 {
-    int (*stepd_connect)(const char *, const char *, uint32_t, uint32_t, uint16_t*);
+    int (*stepd_connect)(const char *, const char *, slurm_step_id_t *, uint16_t*);
     if (setup_libslurm() != SUCCESS) return -1;
     stepd_connect = slurm_stepd_connect;
-    return (*stepd_connect)(dir, hostname, jobid, stepid, protocol);
+    slurm_step_id_t step = { .job_id = jobid,
+                             .step_het_comp = NO_VAL,
+                             .step_id = stepid,};
+    return (*stepd_connect)(dir, hostname, &step, protocol);
 }
 
 int wrap_spank_stepd_add_extern_pid(shifterSpank_config *ssconfig,