Job Configuration
Fine-tune your jobs for optimal performance and reliability.
Accessing Job Settings
- Navigate to Jobs
- Click on the job you want to configure
- Click Edit to modify settings
Configuration Options
Basic Settings
Job Name
The job name cannot be changed after creation. To rename, create a new job and delete the old one.
Group
The group is fixed at creation and cannot be changed afterward. Jobs sharing the same group are displayed together in the list as a collapsible row.
Docker Image
Update the image to deploy a new version:
myregistry.com/my-job:v1.0 → myregistry.com/my-job:v2.0
Changing the image does not affect currently running executions. New executions will use the updated image.
Schedule Settings (CronJob)
Cron Expression
Modify the execution schedule using cron syntax:
| Field | Values | Special Characters |
|---|---|---|
| Minute | 0-59 | *, /, - |
| Hour | 0-23 | *, /, - |
| Day of Month | 1-31 | *, /, -, L, W |
| Month | 1-12 | *, /, - |
| Day of Week | 0-6 (Sun-Sat) | *, /, -, L, # |
Examples:
| Expression | Meaning |
|---|---|
*/15 * * * * | Every 15 minutes |
0 */2 * * * | Every 2 hours |
30 4 * * * | Daily at 4:30 AM |
0 0 * * 0 | Weekly on Sundays |
0 0 1,15 * * | 1st and 15th of month |
Timezone
Select the timezone for schedule interpretation. Common options:
| Timezone | Offset | Use Case |
|---|---|---|
| UTC | +00:00 | Server operations, international |
| Europe/Paris | +01:00/+02:00 | French business hours |
| America/New_York | -05:00/-04:00 | US Eastern business |
| Asia/Tokyo | +09:00 | Japanese business hours |
Use UTC for infrastructure jobs and local timezones only for business-specific tasks.
Concurrency Policy
| Policy | When Previous is Running | Best For |
|---|---|---|
| Allow | Start new execution | Independent jobs |
| Forbid | Skip this execution | Prevent resource contention |
| Replace | Stop previous, start new | Always run latest |
Suspend
Toggle to pause scheduling:
- Suspended: No new executions created
- Active: Normal scheduling resumes
Compute Resources
Compute Class
Changing compute class affects the CPU:RAM ratio:
| Class | Ratio | Effect |
|---|---|---|
| Eco | 1:2 | Budget-friendly, lower performance |
| Standard | 1:4 | Balanced for most workloads |
| Production | 1:4 | Dedicated resources, consistent performance |
| Prod. Compute | 1:2 | Optimized for CPU-bound tasks |
| Prod. Memory | 1:8 | Optimized for memory-bound tasks |
| ARM | 1:4 | Cost-effective ARM architecture |
CPU Allocation
Configure in millicores (mCPU):
| Value | Equivalent |
|---|---|
| 100 mCPU | 0.1 cores |
| 500 mCPU | 0.5 cores |
| 1000 mCPU | 1 core |
| 2000 mCPU | 2 cores |
Minimum: 10 mCPU Maximum: Varies by compute class (up to 15000 mCPU for Standard)
Memory Allocation
Memory is automatically calculated based on CPU and compute class ratio:
Example with Standard class (1:4 ratio):
- 500 mCPU → 2048 MiB
- 1000 mCPU → 4096 MiB
- 2000 mCPU → 8192 MiB
A 15% tolerance is allowed for manual adjustment.
Ephemeral Storage
Temporary storage available during execution:
| Range | Use Case |
|---|---|
| 100-500 MiB | Small scripts, minimal temp files |
| 500-2000 MiB | Data processing, file conversions |
| 2000-10240 MiB | Large file processing, builds |
This storage is cleared when the job completes. For persistent data, use databases or object storage.
Execution Settings
Restart Policy
| Policy | Behavior | Use Case |
|---|---|---|
| OnFailure | Retry container on failure | Most jobs |
| Never | Don't retry, fail immediately | When retries aren't safe |
Completions
Number of successful executions required:
| Value | Behavior |
|---|---|
| 1 | Single execution (default) |
| N | Job completes when N pods succeed |
Indexed Jobs: When completions > 1, each pod receives a unique index via JOB_COMPLETION_INDEX environment variable.
Parallelism
Maximum concurrent pods:
| Value | Behavior |
|---|---|
| 1 | Sequential execution (default) |
| N | Up to N pods run simultaneously |
Example: Processing 100 items with parallelism 10 runs up to 10 pods at once.
Backoff Limit
Maximum retries before marking job as failed:
| Value | Behavior |
|---|---|
| 0 | No retries, fail on first error |
| 6 | Retry up to 6 times (default) |
| 100 | Maximum allowed retries |
Retry delay increases exponentially (backoff).
Timeout
Maximum duration (in seconds) an execution may run before Kubernetes kills it. Leave empty for no time limit. The minimum value is 1 second.
On a CronJob update, send null via the API to clear a previously set timeout.
History Limits (CronJob)
Control how many completed executions are retained:
| Setting | Default | Purpose |
|---|---|---|
| Successful Jobs History | 3 | Debug and audit recent successes |
| Failed Jobs History | 1 | Investigate recent failures |
Keep enough history to debug issues, but not so much that it clutters your view.
Container Command
The container command is configured in Basic Configuration alongside the job name and image. Enter the complete command as you would type it in a terminal:
python script.py
sh -c "echo 'Starting' && python script.py && echo 'Done'"
/app/processor --input=/data --output=/results
Environment & Files
Environment Variables
| Type | Storage | Use Case |
|---|---|---|
| Plain | Visible in UI | Configuration values |
| Secret | Masked in UI | Credentials, API keys |
Common Variables:
| Variable | Description |
|---|---|
DATABASE_URL | Database connection string |
API_KEY | External service authentication |
LOG_LEVEL | Logging verbosity (debug, info, warn, error) |
BATCH_SIZE | Processing batch size |
Security Context
| Setting | Description | Default |
|---|---|---|
| Run As User | Linux user ID | Container default |
| Run As Group | Linux group ID | Container default |
Configuration Best Practices
For Reliability
- Set appropriate backoff limits: Too low fails prematurely, too high delays failure detection
- Use Forbid concurrency: Prevent overlapping executions for stateful jobs
- Configure timeouts: Set a Timeout to limit maximum runtime
For Performance
- Right-size resources: Start small, increase based on metrics
- Use parallelism wisely: More parallelism isn't always faster
- Match compute class: CPU-intensive → Prod. Compute, Memory-intensive → Prod. Memory
For Debugging
- Keep sufficient history: At least 1 failed, 3 successful
- Log meaningfully: Include timestamps and progress indicators
- Use structured logging: JSON format for easier parsing
Service Connections
The Connections tab lets you wire your job to managed services (PostgreSQL, MariaDB, Valkey, NATS). SailOps injects the credentials as environment variables so your job can connect without manual configuration.
- CronJobs: you can add, edit, and delete connections at any time. Changes take effect on the next scheduled execution.
- One-time Jobs: connections are read-only after creation because the pod template is immutable.
See Service Connections for alias conventions, access levels, and the full list of injected variables.
Viewing Configuration
Overview Tab
Shows current configuration summary:
- Job type and status
- Schedule (for CronJobs)
- Resource allocation
- Next/last execution times
Configuration Tab
Detailed view of all settings:
- Image and command
- Environment variables
- Resource limits
- Execution parameters
Connections Tab
Lists active service connections with their target, alias, access level, and creation date. On CronJobs you can manage connections directly from this tab.