This document describes the tables that make up the schema. Tables are grouped into categories, and the purpose of each table is explained.
Schema diagram¶
The schema diagrams are automatically generated as PNG images with Graphviz, and show the links between columns of each table. Here follows the overall schema diagram, while the individual diagrams of each category are available below, together with the table descriptions.
Table list¶
Here is the list of all the tables found in a typical database, grouped by categories.
Execution |
Job tracking |
Logging and monitoring |
|---|---|---|
Pipeline structure |
Resources |
|---|---|
Execution¶
Tables to track the agents operating the eHive system
beekeeper¶
Each row in this table corresponds to a beekeeper process that is running or has run on this pipeline.
| Column | Type | Default value | Description | Index |
|---|---|---|---|---|
| beekeeper_id | INTEGER | not set | unique ID for this beekeeper | primary key |
| meadow_host | VARCHAR(255) | not set | hostname of machine where beekeeper started | key: beekeeper_signature_idx |
| meadow_user | VARCHAR(255) | not set | username under which this beekeeper ran or is running | key: beekeeper_signature_idx |
| process_id | VARCHAR(255) | not set | process_id of the beekeeper | key: beekeeper_signature_idx |
| is_blocked | SMALLINT | 0 | beekeeper will not submit workers if this flag is set | |
| cause_of_death | ENUM:
|
NULL | reason this beekeeper exited | |
| sleep_minutes | REAL | NULL | sleep interval in minutes | |
| analyses_pattern | TEXT | NULL | restricting analyses_pattern, if given | |
| loop_limit | INTEGER | NULL | loop limit if given | |
| loop_until | ENUM:
|
not set | beekeeper’s policy for responding to possibly loop-ending events | |
| options | TEXT | NULL | all options passed to the beekeeper | |
| meadow_signatures | TEXT | NULL | signatures for all meadows this beekeeper can submit to |
role¶
Entries of this table correspond to Role objects of the API. When a Worker specialises, it acquires a Role, which is a temporary link between the Worker and a resource-compatible Analysis.
| Column | Type | Default value | Description | Index |
|---|---|---|---|---|
| role_id | INTEGER | not set | unique ID of the Role | primary key |
| worker_id | INTEGER | not set | the specialised Worker | key: worker |
| analysis_id | INTEGER | not set | the Analysis into which the Worker specialised | key: analysis |
| when_started | TIMESTAMP | CURRENT_TIMESTAMP | when this Role started | |
| when_finished | TIMESTAMP | NULL | when this Role finished. NULL may either indicate it is still running or was killed by an external force. | |
| attempted_jobs | INTEGER | 0 | counter of the number of attempts | |
| done_jobs | INTEGER | 0 | counter of the number of successful attempts |
worker¶
Entries of this table correspond to Worker objects of the API. Workers are created by inserting into this table so that there is only one instance of a Worker object in the database. As Workers live and do work, they update this table, and when they die they update again.
| Column | Type | Default value | Description | Index |
|---|---|---|---|---|
| worker_id | INTEGER | not set | unique ID of the Worker | primary key |
| meadow_type | VARCHAR(255) | not set | type of the Meadow it is running on | key: meadow_process |
| meadow_name | VARCHAR(255) | not set | name of the Meadow it is running on (for “LOCAL” meadows it is the same as meadow_host) | key: meadow_process |
| meadow_host | VARCHAR(255) | NULL | execution host name | |
| meadow_user | VARCHAR(255) | NULL | scheduling/execution user name (within the Meadow) | |
| process_id | VARCHAR(255) | not set | identifies the Worker process on the Meadow (for “LOCAL” is the OS PID) | key: meadow_process |
| resource_class_id | INTEGER | NULL | links to Worker’s resource class | |
| work_done | INTEGER | 0 | how many jobs the Worker has completed successfully | |
| status | VARCHAR(255) | ‘READY’ | current status of the Worker | |
| beekeeper_id | INTEGER | NULL | beekeeper that created this worker | |
| when_submitted | TIMESTAMP | CURRENT_TIMESTAMP | when the Worker was submitted by a Beekeeper | |
| when_born | TIMESTAMP | NULL | when the Worker process was started | |
| when_checked_in | TIMESTAMP | NULL | when the Worker last checked into the database | |
| when_seen | TIMESTAMP | NULL | when the Worker was last seen by the Meadow | |
| when_died | TIMESTAMP | NULL | if defined, when the Worker died (or its premature death was first detected by GC) | |
| cause_of_death | VARCHAR(255) | NULL | if defined, why did the Worker exit (or why it was killed) | |
| log_dir | VARCHAR(255) | NULL | if defined, a filesystem directory where this Worker’s output is logged |
Job tracking¶
Tables to list all the jobs of each analysis, and their dependencies
accu¶
Accumulator for funneled dataflow.
| Column | Type | Default value | Description | Index |
|---|---|---|---|---|
| sending_job_id | INTEGER | NULL | semaphoring job in the “box” | key: accu_sending_idx |
| receiving_semaphore_id | INTEGER | not set | semaphore just outside the “box” | key: accu_receiving_idx |
| struct_name | VARCHAR(255) | not set | name of the structured parameter | |
| key_signature | VARCHAR(255) | not set | locates the part of the structured parameter | |
| value | MEDIUMTEXT | NULL | value of the part |
analysis_data¶
A generic blob-storage hash. Currently the only legitimate use of this table is “overflow” of job.input_ids: when they grow longer than 254 characters the real data is stored in analysis_data instead, and the input_id contains the corresponding analysis_data_id.
| Column | Type | Default value | Description | Index |
|---|---|---|---|---|
| analysis_data_id | INTEGER | not set | primary id | primary key |
| md5sum | CHAR(32) | not set | checksum over the data to quickly detect (potential) collisions | key: md5sum |
| data | MEDIUMTEXT | not set | text blob which holds the data |
job¶
The job table is the heart of this system. It is the kiosk or blackboard where workers find things to do and then post work for other works to do. These jobs are created prior to work being done, are claimed by workers, are updated as the work is done, with a final update on completion.
| Column | Type | Default value | Description | Index |
|---|---|---|---|---|
| job_id | INTEGER | not set | autoincrement id | primary key |
| prev_job_id | INTEGER | NULL | previous job which created this one | |
| analysis_id | INTEGER | not set | the analysis_id needed to accomplish this job. | unique key: input_id_stacks_analysis key: analysis_status_retry |
| input_id | CHAR(255) | not set | input data passed into Analysis:RunnableDB to control the work | unique key: input_id_stacks_analysis |
| param_id_stack | CHAR(64) | ‘’ | a CSV of job_ids whose input_ids contribute to the stack of local variables for the job | unique key: input_id_stacks_analysis |
| accu_id_stack | CHAR(64) | ‘’ | a CSV of job_ids whose accu’s contribute to the stack of local variables for the job | unique key: input_id_stacks_analysis |
| role_id | INTEGER | NULL | links to the Role that claimed this job (NULL means it has never been claimed) | key: role_status |
| status | ENUM:
|
‘READY’ | state the job is in | key: analysis_status_retry key: role_status |
| retry_count | INTEGER | 0 | number times job had to be reset when worker failed to run it | key: analysis_status_retry |
| when_completed | TIMESTAMP | NULL | when the job was completed | |
| runtime_msec | INTEGER | NULL | how long did it take to execute the job (or until the moment it failed) | |
| query_count | INTEGER | NULL | how many SQL queries were run during this job | |
| controlled_semaphore_id | INTEGER | NULL | the dbID of the semaphore that is controlled by this job (and whose counter it will decrement by 1 upon successful completion) |
job_file¶
For testing/debugging purposes both STDOUT and STDERR streams of each Job can be redirected into a separate log file. This table holds filesystem paths to one or both of those files. There is max one entry per job_id and retry.
| Column | Type | Default value | Description | Index |
|---|---|---|---|---|
| job_id | INTEGER | not set | foreign key | primary key |
| retry | INTEGER | not set | copy of retry_count of job as it was run | primary key |
| role_id | INTEGER | not set | links to the Role that claimed this job | key: role |
| stdout_file | VARCHAR(255) | NULL | path to the job’s STDOUT log | |
| stderr_file | VARCHAR(255) | NULL | path to the job’s STDERR log |
semaphore¶
The semaphore table is our primary inter-job dependency relationship. Any job may control up to one semaphore, but the semaphore can be controlled by many jobs. This includes remote jobs, so the semaphore keeps two counters - one for local blockers, one for remote ones. As soon as both counters reach zero, the semaphore unblocks one dependent job - either a local one, or through a chain of dependent remote semaphores.
| Column | Type | Default value | Description | Index |
|---|---|---|---|---|
| semaphore_id | INTEGER | not set | autoincrement id | primary key |
| local_jobs_counter | INTEGER | 0 | the number of local jobs that control this semaphore | |
| remote_jobs_counter | INTEGER | 0 | the number of remote semaphores that control this one | |
| dependent_job_id | INTEGER | NULL | either NULL or points at a local job to be unblocked when the semaphore opens (exclusive with dependent_semaphore_url) | unique key: unique_dependent_job_id |
| dependent_semaphore_url | VARCHAR(255) | NULL | either NULL or points at a remote semaphore to be decremented when this semaphore opens (exclusive with dependent_job_id) |
Logging and monitoring¶
Tables to store messages and feedback from the execution of the pipeline
analysis_stats_monitor¶
A regular timestamped snapshot of the analysis_stats table.
| Column | Type | Default value | Description | Index |
|---|---|---|---|---|
| when_logged | TIMESTAMP | CURRENT_TIMESTAMP | when this snapshot was taken | |
| analysis_id | INTEGER | not set | foreign-keyed to the corresponding analysis_base entry | |
| status | ENUM:
|
‘EMPTY’ | cached state of the Analysis | |
| total_job_count | INTEGER | 0 | total number of Jobs of this Analysis | |
| semaphored_job_count | INTEGER | 0 | number of Jobs of this Analysis that are in SEMAPHORED state | |
| ready_job_count | INTEGER | 0 | number of Jobs of this Analysis that are in READY state | |
| done_job_count | INTEGER | 0 | number of Jobs of this Analysis that are in DONE state | |
| failed_job_count | INTEGER | 0 | number of Jobs of this Analysis that are in FAILED state | |
| num_running_workers | INTEGER | 0 | number of running Workers of this Analysis | |
| avg_msec_per_job | INTEGER | NULL | weighted average | |
| avg_input_msec_per_job | INTEGER | NULL | weighted average | |
| avg_run_msec_per_job | INTEGER | NULL | weighted average | |
| avg_output_msec_per_job | INTEGER | NULL | weighted average | |
| when_updated | TIMESTAMP | NULL | when this entry was last updated | |
| sync_lock | SMALLINT | 0 | a binary lock flag to prevent simultaneous updates | |
| is_excluded | SMALLINT | not set | set to exclude analysis from beekeeper scheduling |
log_message¶
When a Job or a job-less Worker (job_id=NULL) throws a “die” message for any reason, the message is recorded in this table. It may or may not indicate that the job was unsuccessful via is_error flag. Also $self->warning(”…”) messages are recorded with is_error=0.
| Column | Type | Default value | Description | Index |
|---|---|---|---|---|
| log_message_id | INTEGER | not set | an autoincremented primary id of the message | primary key |
| job_id | INTEGER | NULL | the id of the job that threw the message (or NULL if it was outside of a message) | key: job_id |
| role_id | INTEGER | NULL | the “current” role | |
| worker_id | INTEGER | NULL | the “current” worker | key: worker_id |
| beekeeper_id | INTEGER | NULL | beekeeper that generated this message | key: beekeeper_id |
| when_logged | TIMESTAMP | CURRENT_TIMESTAMP | when the message was thrown | |
| retry | INTEGER | NULL | retry_count of the job when the message was thrown (or NULL if no job) | |
| status | VARCHAR(255) | ‘UNKNOWN’ | of the job or worker when the message was thrown | |
| msg | MEDIUMTEXT | NULL | string that contains the message | |
| message_class | ENUM:
|
‘INFO’ | type of message |
worker_resource_usage¶
A table with post-mortem resource usage statistics of a Worker. This table is not automatically populated: you first need to run load_resource_usage.pl. Note that some meadows (like LOCAL) do not support post-mortem inspection of resource usage
| Column | Type | Default value | Description | Index |
|---|---|---|---|---|
| worker_id | INTEGER | not set | links to the worker table | primary key |
| exit_status | VARCHAR(255) | NULL | meadow-dependent, in case of LSF it’s usually “done” (normal) or “exit” (abnormal) | |
| mem_megs | FLOAT | NULL | how much memory the Worker process used | |
| swap_megs | FLOAT | NULL | how much swap the Worker process used | |
| pending_sec | FLOAT | NULL | time spent by the process in the queue before it became a Worker | |
| cpu_sec | FLOAT | NULL | cpu time (in seconds) used by the Worker process. It is often lower than the walltime because of time spent in I/O waits, but it can also be higher if the process is multi-threaded | |
| lifespan_sec | FLOAT | NULL | walltime (in seconds) used by the Worker process. It is often higher than the sum of its jobs’ “runtime_msec” because of the overhead from the Worker itself | |
| exception_status | VARCHAR(255) | NULL | meadow-specific flags, in case of LSF it can be “underrun”, “overrun” or “idle” |
Pipeline structure¶
Tables to store the structure of the pipeline (i.e. the analyses and the dataflow-rules between them)
analysis_base¶
Each Analysis is a node of the pipeline diagram. It acts both as a “class” to which Jobs belong (and inherit from it certain properties) and as a “container” for them (Jobs of an Analysis can be blocking all Jobs of another Analysis).
| Column | Type | Default value | Description | Index |
|---|---|---|---|---|
| analysis_id | INTEGER | not set | a unique ID that is also a foreign key to most of the other tables | primary key |
| logic_name | VARCHAR(255) | not set | the name of the Analysis object | unique key: logic_name_idx |
| module | VARCHAR(255) | not set | the name of the module / package that runs this Analysis | |
| language | VARCHAR(255) | NULL | the language of the module, if not Perl | |
| parameters | MEDIUMTEXT | NULL | a stringified hash of parameters common to all jobs of the Analysis | |
| resource_class_id | INTEGER | not set | link to the resource_class table | |
| failed_job_tolerance | INTEGER | 0 | % of tolerated failed Jobs | |
| max_retry_count | INTEGER | NULL | how many times a job of this Analysis will be retried (unless there is no point) | |
| can_be_empty | SMALLINT | 0 | if TRUE, this Analysis will not be blocking if/while it doesn’t have any jobs | |
| priority | SMALLINT | 0 | an Analysis with higher priority will be more likely chosen on Worker’s specialisation | |
| meadow_type | VARCHAR(255) | NULL | if defined, forces this Analysis to be run only on the given Meadow | |
| analysis_capacity | INTEGER | NULL | if defined, limits the number of Workers of this particular Analysis that are allowed to run in parallel | |
| hive_capacity | INTEGER | NULL | a reciprocal limiter on the number of Workers running at the same time (dependent on Workers of other Analyses) | |
| batch_size | INTEGER | 1 | how many jobs are claimed in one claiming operation before Worker starts executing them | |
| comment | TEXT | not set | human-readable textual description of the analysis | |
| tags | TEXT | not set | machine-readable list of tags useful for grouping analyses together |
analysis_ctrl_rule¶
These rules define a higher level of control. These rules are used to turn whole analysis nodes on/off (READY/BLOCKED). If any of the condition_analyses are not “DONE” the ctrled_analysis is set to BLOCKED. When all conditions become “DONE” then ctrled_analysis is set to READY The workers switch the analysis.status to “WORKING” and “DONE”. But any moment if a condition goes false, the analysis is reset to BLOCKED.
| Column | Type | Default value | Description | Index |
|---|---|---|---|---|
| analysis_ctrl_rule_id | INTEGER | not set | internal ID | primary key |
| condition_analysis_url | VARCHAR(255) | ‘’ | foreign key to net distributed analysis reference | unique: key |
| ctrled_analysis_id | INTEGER | not set | foreign key to analysis table analysis_id | unique: key |
analysis_stats¶
Parallel table to analysis_base which provides high level statistics on the state of an analysis and it’s jobs. Used to provide a fast overview, and to provide final approval of “DONE” which is used by the blocking rules to determine when to unblock other analyses. Also provides
| Column | Type | Default value | Description | Index |
|---|---|---|---|---|
| analysis_id | INTEGER | not set | foreign-keyed to the corresponding analysis_base entry | primary key |
| status | ENUM:
|
‘EMPTY’ | cached state of the Analysis | |
| total_job_count | INTEGER | 0 | total number of Jobs of this Analysis | |
| semaphored_job_count | INTEGER | 0 | number of Jobs of this Analysis that are in SEMAPHORED state | |
| ready_job_count | INTEGER | 0 | number of Jobs of this Analysis that are in READY state | |
| done_job_count | INTEGER | 0 | number of Jobs of this Analysis that are in DONE state | |
| failed_job_count | INTEGER | 0 | number of Jobs of this Analysis that are in FAILED state | |
| num_running_workers | INTEGER | 0 | number of running Workers of this Analysis | |
| avg_msec_per_job | INTEGER | NULL | weighted average | |
| avg_input_msec_per_job | INTEGER | NULL | weighted average | |
| avg_run_msec_per_job | INTEGER | NULL | weighted average | |
| avg_output_msec_per_job | INTEGER | NULL | weighted average | |
| when_updated | TIMESTAMP | NULL | when this entry was last updated | |
| sync_lock | SMALLINT | 0 | a binary lock flag to prevent simultaneous updates | |
| is_excluded | SMALLINT | 0 | set to exclude analysis from beekeeper scheduling |
dataflow_rule¶
Each entry of this table defines a starting point for dataflow (via from_analysis_id and branch_code) to which point a group of dataflow_target entries can be linked. This grouping is used in two ways: (1) dataflow_target entries that link into the same dataflow_rule share the same from_analysis, branch_code and funnel_dataflow_rule (2) to define the conditions for DEFAULT or ELSE case (via excluding all conditions explicitly listed in the group)
| Column | Type | Default value | Description | Index |
|---|---|---|---|---|
| dataflow_rule_id | INTEGER | not set | internal ID | primary key |
| from_analysis_id | INTEGER | not set | foreign key to analysis table analysis_id | |
| branch_code | INTEGER | 1 | branch_code of the fan | |
| funnel_dataflow_rule_id | INTEGER | NULL | dataflow_rule_id of the semaphored funnel (is NULL by default, which means dataflow is not semaphored) |
dataflow_target¶
This table links specific conditions with the target object (Analysis/Table/Accu) and optional input_id_template.
| Column | Type | Default value | Description | Index |
|---|---|---|---|---|
| dataflow_target_id | INTEGER | not set | internal ID | primary key |
| source_dataflow_rule_id | INTEGER | not set | foreign key to the dataflow_rule object that defines grouping (see description of dataflow_rule table) | unique: key |
| on_condition | VARCHAR(255) | NULL | param-substitutable string evaluated at the moment of dataflow event that defines whether or not this case produces any dataflow; NULL means DEFAULT or ELSE | unique: key |
| input_id_template | MEDIUMTEXT | NULL | a template for generating a new input_id (not necessarily a hashref) in this dataflow; if undefined is kept original | unique: key |
| extend_param_stack | SMALLINT | 0 | the boolean value defines whether the newly created jobs will inherit both the parameters and the accu of the prev_job | |
| to_analysis_url | VARCHAR(255) | ‘’ | the URL of the dataflow target object (Analysis/Table/Accu) | unique: key |
hive_meta¶
This table keeps several important hive-specific pipeline-wide key-value pairs such as hive_sql_schema_version, hive_use_triggers and hive_pipeline_name.
| Column | Type | Default value | Description | Index |
|---|---|---|---|---|
| meta_key | VARCHAR(255) | not set | the key of key-value pairs (primary key) | primary key |
| meta_value | MEDIUMTEXT | NULL | the value of key-value pairs |
pipeline_wide_parameters¶
This table contains a simple hash between pipeline_wide_parameter names and their values. The same data used to live in “meta” table until both the schema and the API were finally separated from Ensembl Core.
| Column | Type | Default value | Description | Index |
|---|---|---|---|---|
| param_name | VARCHAR(255) | not set | the key of key-value pairs (primary key) | key: value_idx primary key |
| param_value | MEDIUMTEXT | NULL | the value of key-value pairs |
Resources¶
Tables used to describe the resources needed by each analysis
resource_class¶
Maps between resource_class numeric IDs and unique names.
| Column | Type | Default value | Description | Index |
|---|---|---|---|---|
| resource_class_id | INTEGER | not set | unique ID of the ResourceClass | primary key |
| name | VARCHAR(255) | not set | unique name of the ResourceClass | unique: key |
resource_description¶
Maps (ResourceClass, MeadowType) pair to Meadow-specific resource lines.
| Column | Type | Default value | Description | Index |
|---|---|---|---|---|
| resource_class_id | INTEGER | not set | foreign-keyed to the ResourceClass entry | primary key |
| meadow_type | VARCHAR(255) | not set | if the Worker is about to be executed on the given Meadow… | primary key |
| submission_cmd_args | VARCHAR(255) | ‘’ | … these are the resource arguments (queue, memory,…) to give to the submission command | |
| worker_cmd_args | VARCHAR(255) | ‘’ | … and these are the arguments that are given to the worker command being submitted |